Slightly tangential, but this is the reason I'm baffled why people think that AI-driven podcasts would ever be worth listening to.
If you can find an LLM+TTS generated podcast with even a FRACTION of the infectious energy as something like Good Job Brain, Radio Lab, or Ask Me Another, then I'll eat my hat. I don't even own one. I'll drive to the nearest hat store, purchase the tallest stovetop hat that I can afford and eat it.
I totally agree, but I also think it's just not geared towards us because of our age and preexisting beliefs of "what podcasts/news/videos are supposed to be". Think of kids around age <12. If they'll keep consuming AI-driven media, they'll take that as normal and won't blink twice if that becomes the de-facto standard in about 10 years.
It's the same as short-video format for me. Sure, I can watch some TikToks from time to time, but making them, or continuously getting my news from them? Yeah, that's not gonna fly for me. However, for my niblings (age 8-17) that's basically where they're getting all the "current affairs" from. Microtransaction is probably one of the easiest example as well. 15 years ago, anyone who bought games would laugh at you if you said every game that you paid $80 for would also have endless amount of small items that you can buy for real money. Right now? Well, kids that grew up with Fortnite and Roblox just think that's the norm.
The younger generation loves short-form, to-the-point stuff. Which is the exact opposite of what the current crop of GenAI makes. In a tiktok video, every sentence, every word counts because there's a time limit. If people don't engage with your content in the first 3 seconds, it's worthless. The video linked in another post starts off with 15 seconds of complete fluff. You'd have better engagement if you have a guy opening with BIIIG NEWS!! LABOR DAY BREAKFAST GOES HAYWIRE!!! and hook people.
GenAI is great at generating "stuff", but what makes good content isn't the quantity. What makes good content is when there's nothing left to take away.
I’m only half joking when I’ve described ChatGPT-authored emails as a uniquely inefficient transport format.
Author feeds bullet points into ChatGPT which burns CPU cycles producing paragraphs of fluff. Recipient feeds paragraphs of fluff into ChatGPT and asks it to summarise into bullet points.
Then users turn around and feed the fluff into energy hungry summarizers because who has time for a 5 paragraph email that could’ve been a three point bulleted list?
It would be a net win if it could normalize sending prompts instead of normal communication, which is not far in terms of useless waste of energy and space to LLM output that emulates it.
Similarily, MSFT recently announced the upcoming ability to clone your voice for Teams meetings. Extrapolating, in a few months, there will be Teams meetings which are only frequented by avatars. At the end of the meeting, you get an email with the essential content. Weird times ahead.
The way I explained it when I taught English 101 to first-year university students: any substantive question can generate an answer of a paragraph or a life's work; in this assignment I expect you to go into this much depth. Of course, good expository writing is as to-the-point as possible, so the first hurdle for most students was eliminating the juvenile trick of padding out their prose with waffle to meet an arbitrary word-count. Giving a word-count to an AI seems (currently) to activate the same behavior. I've not yet seen an AI text that's better writing than a college freshman could be expected to produce.
> Of course, good expository writing is as to-the-point as possible, so the first hurdle for most students was eliminating the juvenile trick of padding out their prose with waffle to meet an arbitrary word-count.
This is the most beautiful sentence I’ve read today.
We won't get them unless we appreciate both teaching and the Humanities more than we do. I was good, but by no means the best (75th percentile, maybe?). I loved doing it, but changed careers to IT because I'd never have been able to support a family on what I was paid.
A culture which pays teachers poorly, treats them with disrespect ("those who can do..."), and relentlessly emphasizes STEM, STEM, STEM is one that's slowly killing itself, no matter how much shiny tech at inflated valuations it turns out.
I don't know how it is elsewhere, but where I grew up we had minimum word limits on pretty much all essays. Doesn't matter if you can say what you want to say in 6 sentences, you need 4000 words or 2 pages or whatever metric they use
Oh, of course. Length requirements are important, for the reasons I explained up-thread. However, if teachers accept any old thing ("padding") to reach the count, then that metric is arbitrary, which (justifiably!) makes students cynical.
If a student can say all that they want to say in six sentences then they need to learn more about the topic, and / or think through their opinions more deeply. Teachers who do not take that next step are bad teachers, because they are neither modeling nor encouraging critical thinking.
In some places the majority of teachers are themselves incapable of critical thinking, because those who are leave the profession (or the locale) for the reasons in my comment above.
[Edit to add]: Please note that I say bad teachers, not bad people. Same goes for students / citizens, as well. The ability to think critically is not a determinate of moral worth, and in some ways and some cases might be anti-correlated.
Don't get me started on college admissions essays. Rich kids pay other people to write them. Poor kids don't understand the class-markers they're expected to include. If AI consigns them to the dustbin of history it might be the first unalloyed good that tech ever does.
One of my favorite Mark Twain quotes comes from one of his correspondences: 'My apologies for such a long letter, I hadn't the time to write a short one.'
I never had that requirement outside the first years of school- where it’s more about writing practice than writing actual essays.
After it was always “must be below X pages”
X words is supposed to be a proxy for do enough research that you have something to say with depth. A history of the world in 15 minutes better cover enough ground to be worth 15 minutes - as opposed to 1 minute and then filler words. Of course filler is something everyone who writes such a thing and comes up a few words short does - but you are supposed to go find something more to say.
I eventually flipped from moaning about word count minimums to whining about conference page limits but it took a long, long time- well into grad school. The change came when I finally had something to say.
Write a comment explaining that the ostensibly simple task of writing a dozen or so thank you letters for those socks/etc you received for Christmas can, for some people, be an excruciating task that takes weeks to complete, but with the aid of LLMs can easily be done in an hour.
Sure, it's not just thank you cards though. I once had a job in which my boss assigned me the weekly task of manually emailing an automatically generated report to his boss, and insisted that each email have unique body text amounting to "here's that report you asked for" but stretched into three or four sentences, custom written each week and never repeating. The guy apparently hated to receive automated emails and would supposedly be offended if I copy-pasted the same email every time.
Absolutely senseless work, perfect job for an LLM.
I’d prefer to receive no thank you than to receive an AI written one. One says you don’t care, the other says you don’t care but also want to deceive me.
There's the third case: they care. In which you wouldn't be able to tell whether the card is "genuine" or AI written; the two things aren't even meaningfully different in this scenario.
You can tell if the thank-you card is hand-written. Most people don't have a pen plotter connected to their AI text generator to write thank-you notes.
Emailed or texted "thank you" notes don't count. At all.
Yes, but they could also generate the text and transcribe it onto paper by hand.
For many people, myself included, 90%+ work on things like thank-you notes, greetings, invitations, or some types of e-mails, is in coming up with the right words and phrases. LLMs are a great help here, particularly with breaking through the "blank page syndrome".
It's not that different from looking up examples of holiday greetings on-line, or in a book. And the way I feel about it, if an LLM manages to pick just the right words, it's fine to use them - after all, I'm still providing the input, directing the process, and making a choice which parts of output to use and how. Such text is still "mine", even if 90% of it came from GPT-4.
I guess if someone went through the effort to prompt an LLM for a thank-you card note, and then transcribed that by hand to a card and mailed it, that would count. It's somehow more about knowing that they are making some actual effort to send a personalized thank you than it is about who wrote it.
But honestly I don't think "blank page syndrome" is very common for a thank-you card. We're talking about a few sentences expressing appreciation. You don't really have to over-think it. People who don't send thank-yous are mostly just being lazy.
My financial advisor sends out Christmas cards and Birthday cards. They are pre-printed stock cards. I don't even open them. I should tell him not to waste the money. If he even wrote just one sentence that expressed some personal interest, then they would mean something.
These kinds of messages are on the one hand just pro-forma courtesies, but on the other hand they require that that some personal effort is invested, or else they are meaningless.
There's something painfully ironic and disturbing that the pseudo-Kolmogorov complexity of clickbait content, as judged "identical in quality" by an average human viewer, is arguably less than the length of the clickbait headline itself, and perhaps even less than the embedding vector of said headline!
It's always been this way, it's just rules of polite/corporate culture don't allow to say what you actually mean - you have to hit the style points and flatter the right people the right way, and otherwise pad the text with noise.
If the spread of AI would make it OK to send prompts instead of generated output, all it would do is to finally allow communicating without all the bullshit.
Related, a paradox of PowerPoint: it may suck as communication tool, but at the same time, most communication would be better off if done in bullet points.
Isn't it just a matter of time until AI gets trained to generate attention-grabbing videos? Also, "the first three seconds" isn't exactly the case anymore. There's a push for algorithm to favour videos that are longer than 1+ minutes. Which, to my understanding, is TikTok's way of fighting for YouTube's userbase.
The videos are longer in total length, but you've never seen the average TikTok/Insta user if you think people are letting videos play for more than a few seconds before scrolling onto the next one. This is why movie trailer videos now have a "trailer for the trailer" in the opening seconds with like "THE TRAILER FOR SONIC 3... STARTS NOW" with all of the most attention-grabbing scenes frontloaded.
Similar on IG, which is why a lot of the photographers on my feed will post five or six images, "Which is your favorite, 1-6? Comment below!" because they get the engagement, synthetic or otherwise, of you clicking through each image, and then commenting.
Youtubers also adopted this trick. The best pattern currently is to voice a random question in the middle of the video or present themselves as "I'm not sure about this, please correct me in the comments" as transparent plan to entice people to comment. I even bought it a first few dozen times, but now when every single creator does it, it is kinda tiring.
> Isn't it just a matter of time until AI gets trained to...
blah blah yes "it's a matter or time" for every one of the myriad shortcomings of the technology to be resolved. If you're a true believer everything is "a matter of time". I'll believe it when I see it.
Depends on the technology. It’s hard to look at the progress from December of 2022 till today, and think we won’t go further. Image generation is getting better every day. Parts of the video generation pipelines are also advancing.
the counterpoint is since then we've killed all the easy ways to scale. the datasets can't get bigger because it's already the whole Internet. model sizes can't grow that much because you start running into RAM constraints. efficiency definitely can be improved, but probably not more than 100x on our current architectures.
I'm also pretty aure you'll see it eventually... Consider the possibility that this is both a bubble waiting to pop, as well as the stuff that will shape the future. Kind of like the Internet around the year 2000.
Is that true of tiktoks in general? I feel like a lot of the short form videos out there purposefully bait the watcher and drag on for 3/4 of their runtime.
I don't think this is entirely true. Have you seen these "Internet Story told by TTS Voice-over a Minecraft parkour Video" ( which are what my niblings watching ) ? I noticed a lot of the story is dragged on for over minutes. These are the stories that I read in text in about 5 second.
Short form videos are often hyper focus and to the point, but there are a lot of vertical video contents that are just ( to use the GenZ term ) brain rot like these as well.
Short form video content is absolutely terrible, every time I watch one I have an extremely strong urge to verify whatever I heard in the video, so I just avoid watching them altogether. I’m not sure how anything of value can be 45 seconds long with absolutely no context, outside of comedy.
Novelty grabs people's attention. A system based on the statistical analysis of past content won't do novelty. This seems like a very basic issue to me.
Novelty itself is easy, the hard part is the kind of novelty that is familiar enough to be engaging while also unusual enough to attract all the people bored by the mainstream.
Worse, as people attempt to automate novelty, they will be (and have been) repeatedly thwarted by the fact that the implicit patterns of the automation system themselves become patterns to be learned and recognised… which is why all modern popular music sounds so similar that this video got made 14 years ago: https://www.youtube.com/watch?v=5pidokakU4I
(This is already a thing with GenAI images made by people who just prompt-and-go, though artists using it as a tool can easily do much better).
But go too soon, be too novel, and you're in something like uncanny valley: When Saint-Saëns' Danse macabre was first performed, it was poorly received by violating then-current expectations, now it's considered a masterpiece.
A system that digs out undiscovered mechanisms to drape novelty on based on where the statistical analysis says it’s already been, that would do it though.
If there’s a clear distinction between LLMs looking for patterns in text and LLMs looking for patterns in patterns in text, I’m interested in seeing it better and understanding it.
If you want people to watch your content there's definitely a time limit. I don't mean anything imposed by the platform; I just mean if people aren't interested within 3 seconds, they're scrolling to the next video in their feed.
Seconds - that's all the time I give something. If it looks draggy I will skip ahead to like halfway thru the video and if the person isn't in the middle of explaining something that they have clearly spent half the time extrapolating, that's it. They get no more time.
I often turn on subtitles and watch at 2x speed.
I prefer the transcript 100% of the time to video.
When I was a kid, I loved having breakfast made for me by other people as I couldnt cook. As a small child, one would say I expected it. It doesn't mean when I become an adult that would continue to be the case.
> If people don't engage with your content in the first 3 seconds, it's worthless.
Is that based on a vigorous experience as a content creator on TikTok as a long form content creator or are you going off what you've heard about TikTok, or it's what your feed is full of? (which says more about you/your feed than it does about TikTok)
All of those immediately have something attention-grabbing within the first few seconds: picture of Mars' surface, run map showing funny human shape, text with "Gen Z programmers are crazy" prefacing the anecdote, going immediately into the IT-related rap, guy holding a big and cool-looking stick, "dealing with your 10x coworker" immediately showing the point of the video. The only one that doesn't is the SQL one I guess but that's a very low view count (relatively) on a niche channel.
So thanks for providing a bunch of examples that prove my point, I guess?
none of them are breathless "BIIIG NEWS!! LABOR DAY BREAKFAST GOES HAYWIRE!!!" attention demanding within the first three seconds imo, but, sure, whatever, you're totally right
I've never used tiktok and this post was... enlightening. The ClickUP HR guys are actually pretty funny... but wtf is this from that first channel you posted... lol? https://www.tiktok.com/@luckypassengers/video/74395775119539... - I mean, she's not wrong.
The younger generation has had their brain's development hacked by corporate advertising and can't help themselves but prefer the content they have been all but brainwashed to prefer - this was done deliberately and intentionally and obviously to much detriment to the young of today.
Nobody cares tho bc "they can't pay attention" like it's somehow an entire generation's individual faults they are like that...
The idea that "TikTok is for Gen Z" seems like a very stale meme, although I only have anecdata to back that up.
Microtransactions are way older than 15 years. Wizards of the Coast was selling randomized MtG booster packs in 1993. I'm guessing that the earliest loot boxes for kids were baseball card packs, with very similar psychological purpose to today's game cosmetic collectibles.
> I'm guessing that the earliest loot boxes for kids were baseball card packs
Baseball card packs are an innovation. If you read Peanuts, you'll see that they're referred to as "bubble gum cards", because cards were included as promotional items in packs of gum, a "free toy inside" that was compatible with the size and shape of bubble gum. They moved to dedicated packs of cards when people started buying gum to get the cards.
I totally agree, but it’s just easier for people to accept something if they grew up with it. Sure there will always be people from older generations dabbling with new stuff. But quite a few people refuse to change their behaviour as the age. I wrote them as examples, because it is the biggest contrast I can see in online behaviour between myself and my nephews/nieces plus their circles.
I think the main difference is digital vs physical goods. I know it's minor since a card is just a cheap piece of cardboard, but it's still something tangible (unless the game cosmetics also include a physical item, in which case...I'm dumb).
When WotC first offered digital cards, they had a system whereby you could redeem a complete set for its physical equivalent. At the time this helped convince naysayers they weren’t giving money for nothing. Twenty years later, with their latest online offerings, nobody really cares. I’ve certainly spent more money and had more fun with online Magic, and when I wanted to build my own cube I just got proxies printed anyway. I think the “collectible” part fell away a long time ago and now you’re happy just paying for an experience.
If something isn't digital today it isn't as real.
A physical photo in my hand is sooo limited compared to the same photo in my hand via the screen on my smartphone. Same with a DVD or CD or Disc - all antiquated tech.
"If they'll keep consuming AI-driven media, they'll take that as normal and won't blink twice if that becomes the de-facto standard in about 10 years."
There is a sense in which that is true.
However, we all develop taste, and in a hypothetical world where current AI ends up being the limit for another 10 or 20 years, eventually a lot of people would figure out that there's not as much "there" there as they supposed.
The wild card is that we probably don't live in that world, and it's difficult to guess how good AI is going to get.
Even now, the voice of AI that people are complaining about is just the current default voice, which will probably eventually be looked on about as favorably as bell-bottomed jeans or beehive hairdos. It is driven less by the technology itself than a complicated set of desires around not wanting to give the media nifty soundbites about how mean (or politically incorrect) AI can be, and not wanting to be sued. It's minimal prompt engineering even now to change it a lot. "Make a snappy TikTok video about whatever" is not something the tech is going to struggle with. In fact given the general poverty of the state space I would guess it'll outcompete humans pretty quickly.
"eventually a lot of people would figure out that there's not as much 'there' there as they supposed."
Let's be honest: most of us here know there is more 'there' on the myriad university-press books available free on Anna's Archive, than on HN. The reason we still hang out here is desire for socializing, laziness, or pathological doomscrolling; information density doesn't really factor into our choices.
Personally I don’t think it has anything to do with normalcy.
I don’t consume AI media because it’s not very good.
I watched a lot a bad movies and read a lot of bad books as a kid that I can’t stomach now because I’ve read better books and watched better movies. My guess is that kids today would do the same, assuming AI doesn’t improve.
> they'll take that as normal and won't blink twice if that becomes the de-facto standard in about 10 years.
This is in part my worry, but the other part is that there just won't be much other of an option. I can't be the only one that feels like a lot of information is just... shit. Yeah, there's still YouTube channels I find that are brilliant but they definitely are not fitting to the algorithm and it's clear they're "punished" for this. I think we can say the same about blogs and other places. It is just getting much harder to actually discover the gems.
The result of this is that I watch more and read garbage and have less joy. Maybe I'm just addicted to the routine, but I definitely consume less. I just think it is the problem when you try to make something for everyone; you just end up making things with no substance, no soul. As they say, diversity is the spice of life, and good god, could I go for some actual flavor.
Slightly tangential but TikTok isn't just short-form stuff. It also has normal length content, which is where most current-affairs/analyses would fall, outside of clips.
Similarly (somehow?) for Fortnite, 100% of their microtransactions are cosmetic. It's also free, so it's be akin to telling people you could get AAA quality games for free. The $60+++ group of games make up a vanishingly small chunk of the total market, and are more a relic of 'our' generation. Roblox is a platform, more than a game, so it's its own beast. In general though I think younger gamers have become more demanding, in a good way.
On the other hand, I'm getting some of my news from shorter (5 - 10 min these days, though they used to be shorter) videos with talking maps and war machines...and I'm very much not young.
Honestly sometimes i doubt there's a more damagingly, intentionally overloaded term in modern lexicon than "free to play". Many, many things claim that title while placing if not game play, then socially necessary events, items, costumes, locations, quests, game modes, etc behind a paywall. We need a different term to describe the incredibly predatory behavior and psychology behind most of the current space of "free to play" games. (I'm not saying your favorite game is a lie, just that many games claiming this title effectively are)
> anyone who bought games would laugh at you if you said every game that you paid $80 for would also have endless amount of small items that you can buy for real money.
> kids that grew up with Fortnite and Roblox just think that's the norm.
How so? Fortnite and Roblox don’t cost $80, they are free to play with in game purchases.
They’re F2P, but the social pressure from your friends to buy the next skin, mini game and etc. completely normalizes the behaviour as you grow up. Then you take it as a “usual thing” and don’t bat an eye when a game you buy also has different skins in the game.
At least that was my perception when I played Call of Duty with my younger family members.
Go back to the 1990s (even probably the 2000s?) and ask literally anyone what they think of the idea of people spending huge quantities of their time on this mortal plane watching videos of other people talking about how they do their make up and pick their outfits.
Are those videos "worth watching"? Are videos of people playing video games "worth watching"? Are videos of people opening products and saying out loud the information written on the box and also easily accessible on the public internet "worth watching"?
I'm not happy about these developments, but that isn't a factor of much concern to the people driving and following these trends, it turns out.
It's been a long time since "quality" and "success" have been decoupled. Which is to say - I hope you like the taste of hats.
Bad TV is a lot older than that. It probably seems weird nowadays to watch “I Dream of Jeanie” and “Gilligan’s Island” reruns because that’s what was on TV. Or how about game shows and soap operas? Daytime TV was the worst. But people watched.
Ahh, it's about shipwrecked people that somehow don't endlessly screw each other and still adhere to all the same social norms when they are leaving the idea as they did when they arrived. I think they once made a phone out of a coconut - I watched it before school as a kid, great show! Unbelievably innocent and naive.
I'm pretty sure humans have been finding ways to unproductively waste time for millenium...
I'm not sure how watching lightweight videos on subjects you're curious about is any worse than how people wasted time in the past?
Personally I waste time watching videos on outdoor gear, coffee making equipment and PC hardware. I certainy don't regret it because I had no plans to do anything productive with that time either.
At least in the past people were celebrities for a reason other than the number of followers they had on social. It'd be nice if we could return to a time when people were part of the public discourse because they were good at something (or their parents were rich, sadly).
No dog in this fight, I don't know what exactly you're doing, but I'd cautiously point out that it is, in fact, novel to the internet era to watch rando microcelebrities doing makeup step by step, no matter how long we delay acknowledging that with microquibbles.
I could throw in an example of how I'll watch boring videos of a couple playing with their birds for 90 minutes on Youtube. You can link me to the Wikipedia page on slow TV (via Norway), and it won't erase the simple, boring, straightforward, fact that it is a phenomenon.
I didn’t interpret the original comment to be about micro celebrities, but about people supposedly wasting time today in ways they didn’t beforehand. I agree that micro celebrities are a new phenomenon somewhat (although they are also sort of a return to more regional distribution of fame.) But that wasn’t the point being made.
> At least in the past people were celebrities for a reason other than the number of followers they had on social.
“Other than”? Obviously, as social media didn't exist. “Better than” or “more relevant to the things for which there celebrity status was used to direct attention”? Not particularly.
But you ignored why they have so many followers in the first place. It's because they're entertaining to watch/listen to, which for someone in the entertainment business it's as on the nose as you can possibly get. I think absurd to say they're not good at something because I for sure don't have what it takes to record/edit that many videos all day and be that charismatic all the time.
Ahh, you e touched on a big problem with our society rn - nothing is worth doing.
I've been really thinking about lawns lately and how much time men and women spend maintaining them, how much pride many of them have in such activity... lawns aren't real tho, it's just a personal park nobody ever uses that we all thought we wanted bc rich ppl had them. Front lawns especially, just for looks, nobody normal ever sits on them even.
Case in point is all the people that live in an apartment - they don't do lawns. They might think they want to and some might even enjoy up keeping a lawn but it's not an activity that's "worth it" in fact there are many reasons not have a lawn, it isn't an activity that justifies itself as so many pretend.
Everything is like that. Almost nothing any of us do adds to humanities' general progress or improves our own situations even.
Mowing a lawn and watching TV are incredibly similar activities if you have a nice lawnmower.
I've had very similar thoughts about lawn-cutting, but hadn't related it to technology-related trends!
I couldn't agree more. In a broad sense, it's like we've lost contact with our own lives, communities, and cultures. I don't think those things are "dead", but they mean things now which are totally different than what they meant even only a couple of decades ago, and the people living through it sort of know that and kind of say it out loud sometimes even, but can't really wrap their heads around it at the same time and continue to ape the old behaviours (no offence, anyone, I do it too).
"Traditions" live on in a sort of zombie state... we cut the grass, we present the gift cards. It's an odd and fun(ny) moment to be alive.
I'm not disputing that people waste inordinate amounts of time running out the clock before they shuffle off this mortal coil. (cough every X demograph reacts to Y video cough).
Agreement to eat said head apparel is predicated upon "infectious energy" (i.e. quality) - NOT success. I'll draft up a more officious document later.
Note: I am the sole arbiter of what constitutes quality.
> It's been a long time since "quality" and "success" have been decoupled.
I think so, too.
I guess quality was a property of interest in the old days, because the path e.g. for commercial music was: Maximize profit -> Maximize sales -> Maximize what the target audience likes -> Maximize quality.
For TikTok etc. they bypass the market sales stuff and replace it by an 'algorithm', that optimizes for retention, which is tightly coupled to ad revenue. I imagine the algorithm as a function of many arguments.
Just relying on quality is an inefficient approximation in contrast to that.
Good point. I absolutely abhorred reaction videos until I found a channel that works for me. A guy who reacts to videos about ultra luxury mansions. I find it interesting hearing an expert in the field point out things that most of us wouldn't think of. Plus he's got a great sense of humor.
What blew my mind was that there would be a market to watch other people play video games. You never know what catches on.
> Go back to the 1990s (even probably the 2000s?) and ask literally anyone what they think of the idea of people spending huge quantities of their time on this mortal plane watching videos of other people talking about how they do their make up and pick their outfits.
So like supermarket magazines in video form? I think people would get it.
To me, unboxing videos are documentation, not entertainment. When I need to know exactly what's in a box and how it's packaged, an unboxing video is the only source of that information.
As someone who's struggled to really get into podcasts, I'm convinced that most people who enjoy podcasts don't really actively listen to them, they just like that extra bit of noise in the background while they do something else, with the added bonus that they might just listen at just the right time to pick up some interesting factoid.
Don't get me wrong, I like podcasts, Fall of Civilizations being a personal favorite of mine, it's just that my desire to try and actively listen to them requires that I carve out an hour (or much longer in the case of FoC, worth it) of my free-time to eliminate any distractions and focus. Pretty hard to do when there are so many other things I could be doing with my time besides sitting there trying my best to listen.
Anyway, I bring it up because I'm convinced that the people promoting AI-podcasts are mostly made up of the aforementioned people who just listen to them for noise.
I used to listen to podcasts during my commute, when they took the place of listening to NPR or talk radio. Now that I work from home full time, I listen to them when I'm doing dishes or similar chores, going on walks around the city, and when I'm doing any significant amount of driving. I listen to a combination of comedy/history (love The Dollop), politics, and sci-fi commentary.
For the most part, I'm listening pretty actively, but if I'm just sitting there listening without a fairly mindless activity going on, I'll get distracted pretty quickly and find myself looking at my phone...
I don't get them either since the hosts are typically "regular people" talking about often complicated subjects that they are by no means domain experts in.
I find the format of "Dumdum Host A read some articles about something last night and Dumdum Host B asks questions about it" especially grating, unless it's purely opinion-driven and then I still probably won't care unless I read about the hosts and find out they are probably people with opinions worth listening to.
I'd rather read a book or be with my own thoughts without having them be even more crowded by some randos telling me stuff they think they know
Oh yeah, i definitely hate it when uninformed people try to make podcasts on specialized topics they obviously know nothing about. I went searching for niche physics podcasts (anything besides space stuff please!) And the amount of literal high schoolers and undergraduates attempting to explain things they're years away from even seeing in class is painful, most admit that they know nothing and are literally free-associating over a Wikipedia article. I give them a pass though since they have no listeners and are usually young but...
I exclusively listen to domain expert podcasts. History experts, policy/economics experts, aforementioned physics experts. Literally event i listen to is hosted by at least one person who is a domain expert. They can branch out and interview experts in other domains but there's always an expert. I honestly can't say (besides the high schoolers which i searched out) the last time i've even test listened to a podcast hosted by semi-charismatic article readers.
I'd recommend you search reddit for lists of podcasts in your desired area. App recommendations are all shit and there's many posts out there
> I don't get them either since the hosts are typically "regular people" talking about often complicated subjects that they are by no means domain experts in.
Huh? Most podcasts I listen to feature either experts in the field in question or very skilled journalists compiling the opinions of experts.
> Don't get me wrong, I like podcasts, Fall of Civilizations being a personal favorite of mine, it's just that my desire to try and actively listen to them requires that I carve out an hour (or much longer in the case of FoC, worth it) of my free-time to eliminate any distractions and focus.
I don't see why that should be true. Podcasts are great when coupled with mindless work like mowing the lawn, weeding the garden, stacking firewood, walking to the shop, driving etc. You can get virtually 100% out of a podcast while performing tasks like this.
I figured people listened to them while commuting. It seemed the best fit: lots of time, mostly little attention needed (except when you alight or are driving in adverse conditions).
I don't understand enjoying podcasts while just using them for background noise. Maybe because I actually listen, and most podcasts I listen to seemed to have an engaged audience. But then I like listening to people talk about philosophy, politics, sports, true crime, science, history.
I agree. I think I listen to podcasts just for a bit of "social noise" when I'm doing something by myself and for occasionally picking up on something new I hadn't heard of before. For pure information content, I think they're actually very poor. It's not unlike listening to an old AM radio talk show. Hosts repeat themselves, engage in banter, and often oversimplify topics for the sake of a narrative.
I also think AI podcasts could become popular one day for people who just want some background noise and bits of trivia every once in a while. I would argue that a lot of YouTube channels I sometimes have in the background just summarize Wikipedia articles and don't have much of a personal touch anyway, so an AI could do the same thing.
I switched to improvised pods for this reason. Podcasts are for doing the dishes and mowing the lawn and playing factorio. I don't turn on anything with 'meat' unless I have a long car trip ahead of me. I just am not able to sit still and only listen without feeling like I could be doing something else, and when I'm doing something else I'm gonna start tuning out eventually.
Hey Riddle Riddle, Hello from the magic tavern, Artists on artists on artists on artists are my picks for now.
There is nothing you are supposed to be doing with free time, hence free time.
Tbh tho, there is nothing that we are supposed to be doing in this world rn anyways, nobody was born to work any of these jobs, we didn't evolve to flip burgers.
This is the old world, don't place too much stake in it, everything will change and then what?
If you want to listen to a podcast, do it. If you want to do something else, do that.
Don't do what you don't want to do, that's all there is too worry about.
This is partially true for me, but when I listen to podcasts it’s generally in what would otherwise be dead time for me such as when grocery shopping or exercising.
Having this strong of an opinion on something you admittedly don't understand (the appeal of podcasts) is not a great choice. Maybe ask people instead of making up stuff.
I have very strong opinions on lots of things I can't understand why people like doing, I don't need to understand someone's satisfaction before I determine how I feel about what they are driving satisfaction from. That's ridiculous.
I think people can't be alone with their own minds so they fill every second of there quiet time with someone else's thoughts rather than their own.
> I'm convinced that most people who enjoy podcasts don't really actively listen to them,
Or you know not everyone is the same. Just because you struggle with something doesn’t mean that it is not easy and effortless for others.
I can’t dribble a basketball and walk at the same time, still I won’t make the claim that everyone who claims to enjoy playing basketball is somehow fudging it.
I have probably written this comment about a dozen times on HN already, but: I agree completely, because people don't listen to podcasts purely for information, they listen for information plus community, personality, or just a basic human connection.
If you're a content creator today, the best thing you can do to "AI-proof" your work is to inject your personality into it as much as possible. Preferably your physical personality, on video. The future of human content is being as human as possible. AI isn't going to replace that in our lifetimes, if ever.
> The future of human content is being as human as possible. AI isn't going to replace that in our lifetimes, if ever.
I have been working on machine learning algorithms for a long time. Since the time when telling someone I worked on AI was a conversation killer, even with technical people.
AI's are going to understand people better than people understand people. That could be in five years, maybe - many things are happening faster than expected. Or in 15 years. But that might be the outside range.
There is something about human psychology where the faster something changes, the less we are aware of the rate of change. We don't review the steps and their increasing rate that happened before we cared about that tech. We just accept the new thing that finally gets our attention like it was a one off event, instead of an accelerating compounding flood, and imagine it isn't really going to change much soon.
--
I know this isn't a popular view.
But what is happening is on the order of the transition to the first multi-cellular creatures, or the first bodies with specialized cells, the first nervous systems, the first brains, the first creatures to use language. Far bigger than advances such as writing or even the Internet. This is a transition to a completely new mode for the substrate of life and intelligence. The lack of dependency on any particular substrate.
"We", the new generation of things we are building, will have none of the limits and inefficiencies of our ancient slow DNA-style life. Or our stark physical bottlenecks on action, communication, or scaling, and our inability to understand or directly edit our own internal traits.
We will have to face many challenges in the coming years. It can't hurt to mindfully face them earlier than later.
What machine learning algorithm have you worked on that leads you to believe they are capable of having a rich internal cognitive representation anywhere close to any sentient conscious animal?
If you pick any well performing AI architecture, what would lead you to believe that they are not capable of having a rich internal cognitive representation?
The Transformer, well... transforms, at each layer to produce a different representation of the context. What is this but an internal representation? One cannot assess whether that is rich or cognitive without some agreement of what those terms might mean.
LLMs can seemingly convert a variety of languages into an internal representation that encompasses the gist of any of them. This would at least provide a decent argument that the internal representation is 'rich'
As for cognitive? What assessment would you have in mind that would clearly disqualify something as a non-cognitive entity?
I think most people working in this field who are confident feel that they can extend what they know now to make something that looks like a duck, walks like a duck, and quacks like a duck. If that is achieved, on what basis does anyone have to say "But it's not really a duck"?
I'm ok with people saying AI will be never able to perform that well because it doesn't have X, as long as they accept that if it does, one day, perform that well they accept that either X is present, or that X is not relevant.
If you think we're only our observable behaviors or that is the only relevant thing to you then I don't think it's worth getting into this argument. Consider this excerpt from https://scottaaronson.blog/?p=7094#comment-1947377
> Most animals are goal-directed, intentional, sensory-motor agents who grow interior representations of their environments during their lifetime which enables them to successfully navigate their environments. They are responsive to reasons their environments affords for action, because they can reason from their desires and beliefs towards actions.
In addition, animals like people, have complex representational abilities where we can reify the sensory-motor “concepts” which we develop as “abstract concepts” and give them symbolic representations which can then be communicated. We communicate because we have the capacity to form such representations, translate them symbolically, and use those symbols “on the right occasions” when we have the relevant mental states.
(Discrete mathematicians seem to have imparted a magical property to these symbols that *in them* is everything… no, when I use words its to represent my interior states… the words are symptoms, their patterns are coincidental and useful, but not where anything important lies).
In other words, we say “I like ice-cream” because: we are able to like things (desire, preference), we have tasted ice-cream, we have reflected on our preferences (via a capacity for self-modelling and self-directed emotional awareness), and so on. And when we say, “I like ice-cream” it’s *because* all of those things come together in radically complex ways to actually put us in a position to speak truthfully about ourselves. We really do like ice-cream.
> And when we say, “I like ice-cream” it’s because all of those things come together in radically complex ways to actually put us in a position to speak truthfully about ourselves. We really do like ice-cream.
Ok, now prove this is true. Can you do so without invoking unobservable properties? If you can, then observable is all that matters, if you cannot then you have no proof.
Do I seriously have to prove to you that you like ice cream? Have you tried it? If you sincerely believe you are a husk whose language generation is equivalent to some linear algebra then why even engage in a conversation with me? Why should I waste my time proving to you a human that you have a human experience if you don’t believe it yourself?
You don't need to prove to me that I like ice cream. You need to prove to me that you like ice cream. That you even have the capacity to like. Asserting that you have those experiences proves nothing since even a simple basic program 10 print "I like Ice Cream" can do that.
How can you reliably deny the presence of an experience of another if you cannot prove that experience in yourself?
I actually don’t need to prove to you that I’m more than a BASIC program. I mean listen to yourself. You simply don’t live in the real world. If your mom died and we replaced her with a program that printed a bunch of statements that were designed to as closely mimic your conversations with her as much as possible you wouldn’t argue hey this program is just like my mom. But hey maybe you wouldn’t be able to tell the difference behind the curtain so actually it might as well asbe the same thing in your view, right? I mean who are we to deny that mombot is just like your mom via an emergent pattern somewhere deep inside the matrices in an unprovable way /s. Just because I can’t solve the philosophical zombie problem for you at your whim to your rigor doesn’t mean a chatbot has some equivalent internal experience.
I'm not claiming that any particular chatbot has an equivalent experience, I'm claiming there is no basis beyond its behaviour that it does not.
With the duplicate mother problem, if you cannot tell then there is no reason to believe that it is not a being of equivalent nature. That is not the same as identity, for a layman approach to that viewpoint, see Star Trek: TNG, Season 6, Episode 24. A duplicate Will Riker is created but is still a distinct entity (and one might argue, more original since has been transported one fewer times). Acting the same as is not the same as being the same entity. Nevertheless it has no bearing on whether the duplicate is a valid entity in its own right.
You feel like I'm not living in the real world, but I am the one asking what basis we have for knowing things. You are relying on the presumption that the world reflects what you believe it to be. Epistemology is all about idetifying exactly how much we know about the world.
Ok you can have your radically skeptic hard materialist rhetoric. I just don’t take it seriously, and I don’t think you do either. It’s like those people who insist there is no free will and yet go about their day clearly making choices and exercising their free will. If you want to say technically everyone might as well be a philosophical zombie just reacting to things and your internal subjective experience is an illusory phenomenon, fine you can say that as much as you want. In turn I’ll just give up here because you don’t even have a mind that could be changed. I can sit here and claim you’re the equivalent of a void that repeats radically skeptic lines at me. Maybe a sophisticated chatbot or program. Or maybe you’re actually a hallucination since I can’t prove anything really exists outside of my sense. In which case I’m really wasting my time here.
Well I'm a combatibilist, So I certainly believe in free will. I also will accept any entity that consistently acts as if it has a will that it does actually have that. That's has always been my point, treating things as what they appear to be is the only rational approach when you cannot prove or disprove the existence of the property in question.
It follows from that that you cannot exclude something that appears to have a property if you cannot prove it doesn't or even prove it if it does.
Fair enough. I wouldn’t say a program is acting with a will of its own just because it’s trained to respond to questions in a human like way. That doesn't even say anything about its capacity to have a will. Language is a tool that can convey internal state, not the thing itself.
This is basically the Turing test, and like the Turing test it undervalues other elements which can allow for differentiation between “real” and “fake” things. For example - if we can determine that a thing that looks, walks, and quacks like a duck, but doesn’t have the biological heritage markers (that we can easily determine) then it won’t be treated as equivalent to a duck. The social desire to differentiate between real and fake exists and is easily implementable.
In other words: if AIs/robots ever become so advanced that they look, walk, and talk like people, I expect there to be a system which easily determines if the subject has a biological origin or not.
This is way down the line, but in a closer future this will probably just look like verifying someone’s real world identity as a part of the social media account creation process. The alternative is that billion dollar corporations like Meta or YouTube just let their platforms become overrun with AI slop. I don’t expect them to sit on their hands and do nothing.
> AI's are going to understand people better than people understand people.
Maybe, but very little of the “data” that humans use to build their understanding of humans is recorded and available. If it were it’s not obvious it would be economical to train on. If it were economical it’s not obvious that current techniques would actually work that well and by definition no future techniques are known to work yet. I’m not inclined to say it will never happen but there are a few reasons to predict it’ll prove to be significantly harder to build AI that gets out of the uncanny valley that it’s currently in.
You are describing the current state of AI as if it were a stable point.
AI today is far ahead of two years ago. Every year for many years before that, deep learning models broke benchmark after benchmark before that breakout.
There is no indication of any slow down. The complete reverse - we are seeing dramatic acceleration of an already fast moving field.
Both research and resources are pouring into major improvements in multi-modal learning and learning via other means than human data. Such as reinforcement learning, competitive learning, interacting with systems they need to understand via simulated environments and directly.
> You are describing the current state of AI as if it were a stable point.
No I’m not, I’m just not assuming that the S curve doesn’t exist. There’s no guarantee that research results in X orders of magnitude of improvement that will result in AI being better at understanding humanity than humans in the 5 to 15 year timeframe. There’s no guarantee that compute will continue to consistently grow in volume and lower in price, and a few geopolitical reasons why it might become rarer and prohibitively expensive for some time. There’s no reason to assume capital will stay as available to further both AI techniques and compute resources should there be any sign that investments might not eventually pay off. There’s also no reason to assume the global regulatory environment will remain amenable to rapid AI development. Maybe the industry threads all these needles, but there’s good reason to predict it’ll prove doesn’t.
They are not using AI correctly to create such models. I'm not sure I want AGI right away or even at all, so I'm keeping my epiphany close for now but in the current field of AI nothing er wany will come of this bc it's not the right way.
As soon as the incredibly obvious, far too obvious realization is had, AI will make huge, tremendous leaps overnight essentially. Til then, these are just machine like software, the best we've ever made but nothing more than that.
This makes me think about attention span. Scenes in movies, sound bites, everything has been getting shorter over the decades. I know mine has gotten shorter, because I now watch highly edited and produced videos at 2x speed. Sometimes when I watch at 1x speed I find myself thinking "why does this person speak so slowly?"
Algorithmic content is likely to be even more densely packed with stimuli. At some point, will we find ourselves unable to attend to content produced by a human being because algorithmic content has wrecked our attention span?
Sorry but I don’t think this is much evidence of anything. The point at which an AI can imitate a live-streaming human being is decades away. By then, we will almost certainly have developed a “real Human ID” system that verifies one’s humanity. I wrote about this more here:
The idea that AI is just going to eat all human creative activity because technology accelerates quickly is not a real argument, nor does it stand up to any serious projections of the future.
AI is already eating their way up the creative ladder, this is 100% irrefutable. Interns, junior artists, junior developers, etc are all losing jobs to AI now.
The main problem for AI is it doesn't have a coherent creative direction or real output consistency. The second problem is that creativity thrives on novelty but AI likes to output common things. The first is solvable, probably within 5 years, and is going to hollow out creative departments everywhere. The second is effectively unsolvable, though you might find algorithms that mask it temporarily (I'm not sure if this is any different than what humans do).
We're going to end up with "rock star" teams of creative leads who have more agility in discovering novelty and curating aesthetics than AI models. They'll work with a small department comprised of a mix of AI wranglers and artisans that can put manual finishing touches on AI generated output. Overall creative department sizes are probably going to shrink to 20% of current levels but output will increase by 200%+.
How can you both think AI will do a soulless garbage job and that it will displace all the creative people who put their blood sweat and tears into doing art.
If you think it can be 20 times easier to make a movie, then it seems to me that it would be 20 times less impactful to make a creative work, since the market should quickly react to creative success by making a ton of cheap knockoffs of your work until the dead horse is so thoroughly beaten that it's no longer even worth paying an AI to spit out cheap knock-offs
> How can you both think AI will do a soulless garbage job and that it will displace all the creative people who put their blood sweat and tears into doing art.
They don't, that's why they're saying 20% of creative departments will remain. The part that will go is the part that's already soulless - making ads in the style of the current trendy drama series or what have you.
> If you think it can be 20 times easier to make a movie, then it seems to me that it would be 20 times less impactful to make a creative work, since the market should quickly react to creative success by making a ton of cheap knockoffs of your work until the dead horse is so thoroughly beaten that it's no longer even worth paying an AI to spit out cheap knock-offs
That seems pretty backwards, given Jevons paradox. The ease of writing knock-off fanfiction didn't mean people stopped writing novels. The average novel probably has a lot less impact now than in the past, but the big hits are bigger than ever.
Of course it’s possible I’m wrong. But if we make any sort of projection based on current developments, it would certainly seem that live-streaming AI indistinguishable from a human being is vastly beyond the capabilities of anything out today, and given current expenses and development times, seems to be at least a few decades in the future. To me that is an optimistic assumption, especially assuming that live presence or video quality will continue to improve as well (making it harder to fake.)
If you have a projection that says otherwise, I’d be glad to hear it. But if you don’t, then this idea is merely science fiction.
Making predictions about the future that are based on current accelerating developments are how you get people in the 1930s predicting flying cars by 2000.
You imply that the technological development will stop, but that's not what happened to flying cars - they do and could exist. Since the 1930s the technology didn't stop developing - aircraft went jet powered, supersonic, to the edge of space, huge, light, heavy, more efficient, more affordable, safer; cars went faster, more reliable, safer, more efficient, bigger, smaller, more capable, self driving; fuel got more pure, engines got more power per kilogram, computer aided design and modelling of airflow and stress patterns was developed, stronger lighter materials were developed; flying cars have been built:
Moller SkyCar, of course, which never got to free flight, but was built and could fly.
The problems are regulatory, cost, safety, massive amounts of human skill needed, the infrastructure needed, the lack of demand, etc. A Cessna 182 weighs <900 Kg, a Toyota Corolla weighs >1400 Kg and has no wings, no tail. But if we collectively wanted VTOL flying cars enough to put a SpaceX level of talent and money at them, and were willing to pay more, work harder, for less comfort and more maintenance, we could have them. Bit like Hyperloop; a low-pressure tunnel with carriages rushing through it is not impossible, but it's got more failure modes, more cost, more design difficulties and almost no benefits over high speed train and maglev.
I didn't mean to imply that, but that's my fault for just quickly using it as an example.
I do think there will be some slowdown of the technological development, but I think the situation will be similar to flying cars: regulations, social behavior, etc. will prevent AI from simply devouring everything. Specifically, I expect there to be a system which verifies humanity, and thus popular content will ultimately end up being verified-as-real content.
I don't think regulations and social behaviour could stop a Skynet style "AI go FOOM" future. It might be able to stop LLM filler from covering the internet, but given how it can't stop carbon dioxide filler from the atmosphere and "AI text" is less measurable, less easy to stop, and has less clear consequences, I'm not sure I'd bet on it.
Possibly in the sense of going to a small server where ID is verified and hiding away on Discord from the rest of the internet. That would be more like building a bunker to hide from the flying cars, rather than flying cars never happening.
For many people current AI created videos are already confused for real videos and vice versa.
When you follow up on technology by browsing HN, and see the latest advancements, its easier to see or hear the differences, because you know at what to look.
If I see on tv some badly encoded video, especially in fog or water surfaces, it immediately stands out, because I was working with video decoding during the time it was of much less quality. Most people will not notice.
People are judging AI by what abusers of AI put out there for lols and what they themselves can wring out of it. They haven't yet seen what a bunch of AAA professionals with nothing to lose can build and align.
Billions of dollars and all of Silicon Valley's focus has been spent over the last 2 years trying to get AI to work, the "AAA professionals" are already working on AI and I still have yet to see an AI generated product that's interesting or compelling.
The cornerstone of genAI hype is "AI for thee, not for me"
it's filled with tech people who are fucking morons thinking that everyone else is really dumb and loves slop and their algorithm will generate infinite content for the masses, making them rich. yet they don't consume it themselves, and aren't smart enough to recognize the cognitive dissonance
anyone who thinks unsupervised AI content is going to replace [insert creative output here] shouldn't be in the HN comment section, they should be using ChatGPT to generate comments they can engage with
instead they're going to get mad about me calling them a fucking moron in this comment. which, like, why get mad, you can go get an LLM to generate a comment that's much nicer and agrees with you
> anyone who thinks unsupervised AI content is going to replace [insert creative output here] shouldn't be in the HN comment section, they should be using ChatGPT to generate comments they can engage with
Actually, their handlers should just be better monitoring their internet usage
uh yes? Was this supposed to be some clever "gotcha!" that ended up being really stupid? Might want to run your clever comments by ChatGPT before you post next time chief
oh word? i know a guy at Microsoft who doesn't use Excel, that means nobody at Microsoft ever uses Excel and they're just peddling crap to the dumb masses
the fact that you're engaging w/ a random human on the internet proves that my argument is strong
if you don't like it, go talk to Claude about it
Claude will even use proper punctuation & grammar which i'm not doing. it's LITERALLY an objective improvement over this comment. why the hell are you engaging with this crap?
No AAA professionals in the entertainment profession have built a gen AI model up through their own vision that I'm aware of. AAA Nerds build the next version to distract away from flaws of the last and hope the api will get heavy use. I wouldn't expect output to be compelling with that attitude. I expect repulsive or parody clips to continue until some great creatives feel like building their replacements properly.
Yes, this seems possible. I only wonder if it is too fragile to be self perpetuating. But we're here after all, and this place used to be just a wet rock.
I just took a poll about art that had 50 examples and you had to choose which were human and which were not - I did very well, bc I'm awesome but apparently it shook the art world to its core, many of the professionals have no idea how to tell the difference/can't tell the difference.
The elements that identified a painting as human to me had nothing to do with appearance but everything to do with feeling.
Be passionate about what you present - if you can't be, don't bother. Passion is what will distinguish us from the AI - stuff like personality is almost immediately replicated, vocal inflections, funny comments, unusual spoken delivery - that's all within the realm of AI capability. Talking about micro blogging like it's God's gift to the world, not so much an AI thing, it won't feel right with AI.
Agreed. I finally got around to listening to a NotebookLM generated podcast this weekend, and found it absolutely unlistenable.
For some reason the LLM seemed to latch on to the idea that one host should say something, and the other host should just repeat a key phrase in the sentence for emphasis -- and say nothing else, and they'd do it like multiple times in a sentence, over and over again throughout the entire thing.
Slightly less weird -- but it seems like the LLM caught on that a good narrative structure for a two-host podcast is that one host is the 'expert' on the topic, and the other host plays dumb and ask questions. Not an unreasonable narrative structure. Except that the hosts would seamlessly, and very weirdly, switch roles constantly throughout the podcast.
And ultimately the result was just a high level summary of the article I had provided. They told me in the intro and the outro about the interesting parts they were going to dive into, but they never actually got around to diving into those parts.
I tried it with something fairly abstract about a decision I was working on making. Fed it a bunch of information, details about me, my background, the factors I'm considering in the decision and the impact of getting it right/wrong.
It was interesting. I certainly wouldn't say it was useless, I think the contrived dialogue actually touched on some angles I hadn't considered and I think it was useful. Not in a 'oh shit it's clear to me now' kind of way but it definitely advanced my thinking.
I find it as listenable as the podcasts the people I'm around play in their cars during roadtrips
To me that's absolutely unlistenable, but to them its interesting and engaging. I find NotebookLM replicates that perfectly. Its not at all the issue that the OP encountered with the Hawaii news service, as those lacked tone and pronounciation, which NotebookLM would not.
Regarding the outcome being a long winded summary, yeah thats what I see about the aforementioned 2 hour podcasts as well. They take the intro paragraph of a wikipedia article, pretend that the topic is a novel mystery they just discovered through hours of scouring microfilm at a municipal library, and then interject each other every other word with nonsense, before getting back to the point for just one sentence.
> For some reason the LLM seemed to latch on to the idea that one host should say something, and the other host should just repeat a key phrase in the sentence for emphasis -- and say nothing else, and they'd do it like multiple times in a sentence, over and over again throughout the entire thing.
> If you can find an LLM+TTS generated podcast with even a FRACTION of the infectious energy
this is the wrong standard. there's a long tail of many many many many topics that a Radio Lab will never have the resources or critical mass or staff or turnaround time to cover. AI can serve the long tail like nothing you have ever seen. (because humans are expensive, and deservedly so, but our needs and wants are infinite)
You're not wrong, but your argument seems to boil down to "something is better than nothing" and I suppose I disagree. If there's not a podcast about a particular topic that I'm interested in, I would rather go find a curated expose of that topic via a different medium, such as longform journalism.
I think you make a good connection with the long tail. That will definitely be one of AI's triumphs. People who don't understand this, don't understand what's been happening to media production and consumption over the last 40 years.
Infectious energy is something I expect faked relatively easily, though I don't know your examples and doing it in AI might be a "first 90%" situation just like self driving cars; for me the problem is that they're fairly mediocre at the actual script — based on me putting a blog post I wrote into one and listening to what came out.
Given how many podcasts exist, I think you need to be at least 2 standard deviations above mean to even get noticed, 3 to be a moderate success, and 4 to be in the charts.
I'd guess AI is "good enough" to be 1 above average, as the NotebookLM voices sound like people speaking clearly and with some joy into decent microphones in sound isolating studios.
I probably should've clarified that by infectious energy I wasn't so much referring to the vocal aspect as I was the overall quality, interaction between the hosts, and pithiness / wit.
Having experimented with many LLMs (mixtral, sonnet, ChatGPT, Llama, etc.), the coherence is for the most part on point, but their capacity for novelty has been found wanting irrespective of how I tuned the top_k, temperature, or prompts.
That being said, I've seen some very impressive examples of style transference even conveying emotional range in some of the SOTA TTS systems.
Excuse me for being an old geezer but at least AI bots don't tend to pepper their sentences with frequent utterances of 'like'. I don't normally find this speech mannerism annoying, and I do it myself too, but when 'like' is overused I switch off.
For this reason I don't listen to the otherwise highly entertaining CineFix podcast. Example: the recent episode discussing Kill Bill vol 1 contains 624 utterances in 75 mins (8.3 per min). IANAL I know.
From 2020 until now, we've gone from crude blurry or clearly generative artefacts to being able to create full professional illustrations based upon textual prompts. That is huge. Classic generative art techniques look like cave paintings compared to what the latest image generation models put out (and I'm not talking about "AI slop" type stuff that DALL-E does).
Similarly, tools could fabricate podcasts years ago that sounded terrible. Now we have NotebookLM doing a "reasonable" job with two cliched-sounding "hosts". In a few years, will they potentially be able to create something akin to a professionally produced podcast given some smart prompting? The progress made so far points to yes, and I haven't seen any evidence so far to be pessimistic about it happening.
> "being able to create full professional illustrations based upon textual prompts. That is huge. .. (and I'm not talking about "AI slop" type stuff that DALL-E does)"
Then what are you talking about, where can I get it?
I've heard so many people say that podcasts are just something they have playing in the background while they do other things that I have no trouble believing that they'd play an AI podcast in the same way.
I personally would not bother with AI generated podcasts, they're such low bar, why waste time where there's so much other great content to catch up on? But I think you may be right, I wouldn't be surprised if people take them in with no fuss. But then what do I care? What I care most is that they'll pollute the search space. I'd filter out all GenAI content if I had the option and Im guessing that will become an option soon.
The value in AI podcasts at the moment isn't in replacing human content, but in filling the niches that human content just doesn't cover. Doesn't matter if it's not the best podcast ever, when it's literally the only podcast on this planet discussing the topic you want to listen too.
People think they want to use the Minority Report computer interface because it looks cool and advanced but they don't put the least amount of thought into it and realize it's terribly impractical. Our arms would get tired very quickly. A mouse resting on a table isn't further from ideal just because it was invented earlier.
Fooling people with the promises of AI is pretty simple. People are easy to fool. They like shiny objects.
When TV was first introduced, the first broadcasts were people standing by a mic reading a radio script.
AI podcasts aren't going to be drop in replacements for exactly the length, frequency, personalities, topics, and style of current podcasts. Claiming that's a fault with AI podcasts just indicates a lack of imagination.
The main reason im not concerned about AI-based entertainment is the same reason I watch human chess players. It's not only about technical capabilities. I can't explain fully why though..
I've heard amazing catchy songs from Suno and Udio. So much so they're still stuck in my head as earworms several months later. If they'd been streaming on youtube or spotify I wouldn't have given it a thought that they might be AI generated.
So, I can certainly imagine a podcast doing the same to some degree. Maybe not a podcast where AI wrote the script, but, a podcast where AI read a story dramatically doesn't seem too far off or, easier, a podcast that read news to me.
Again - think of where we were two years ago. I never understand this hubris people have to think AI can never do X while being proved repeatedly wrong.
GPT styled LLMs were introduced back in 2018 so SIX years ago.
Have they gotten more COHERENT? Absolutely. Is coherence the same thing as NOVELTY? NOT EVEN REMOTELY. I've played with markov chains in the 90s that were capable of producing surprising content.
Unless there is a radical advancement in the underlying tech, I don't see any indication that they'll be capable of genuine novelty any time in the near future.
Take satire for example. I have yet to see anything come out of an LLM that felt particularly funny. I suppose if the height of humor for you is Dad jokes, reddit level word punnery, and the backs of snapple lids though that might be different.
If you have a particular style of witty observational humor that you prefer, providing the model some examples of that will help it generate better output. It's capable of generating pretty much anything if you prompt it the right way. For truly nuanced or novel things, you have to give it a nucleus of novelty and axis of nuance for it to start generating things in your desired space.
I totally agree - LLMs may be great for spitting out content, but they just can’t quite replicate the energy, personality, and creativity of an actual podcast. The human touch is the irreplaceable component.
It doesn't quite stop at podcast hosts either. The simple fact that LLMs will never truly be capable of replacing software developers was one of the core principles behind our team developing The Ac28R - A completely new type of AI capable of bulletproof industrial software. It won’t host a podcast anytime soon, but if you need flawless code for complex projects, it's got you covered!
I think we will get a surprising amount of AI generated content in the future. During the first year of the Urkain invasion there was an enormous amount of AI voiced and scripted video content on YouTube. I think AI content will take over in the easy parts first. And over time take up more and more views.
I only discovered https://notebooklm.google.com/ today, as an experiment i threw in a dry EU directive, https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CEL... and got it to generate a podcast that was 20 mins long and was a good intro to the directive and natural to listen to. It won't replace podcasts that require creativity but as a tool to summarize info into a digestible format I'll most definitely be using it again.
Businesses/CEOs want to show profitability by spending less on human employees. Human consumers don't want to lose the human touch. Will be really interesting to see how many of the consumer facing AI startups actually make it.
I've never found anyone telling me their opinions to be infectious - I can't even imagine using that word to describe a podcast tbh. Its a form of media I've never understood - they script the podcasts, they all kno the conversation ahead of time and just go thru the motions via voice with strangers that can either agree or disagree - only the extremes will ever interact with the podcasters, it's just so fake I can't even do it.
I do enjoy some of the podcasts mentioned in this thread, but struggle to find good non-American ("foreign"?!) podcasts to listen to. Similarly to finding good quality non-US film and television, it can be hard to locate but I greatly enjoy it.
If anyone has suggestions for podcast collation sites that are for non American content that would be fantastic.
same is true for AI music, AI art, AI articles to some extent. most communication is about human connection. you remove it and the communication becomes worthless.
AI is great for communication that has less personal involvement like drafting short professional emails. that's also why all the AI agent craze seems problematic: https://medium.com/thoughts-on-machine-learning/langchains-s...
I’d rather listen to silence than a podcast or FM/AM talk radio. I’ve never understood the appeal. In fact music is something I listen to while traveling because it helps me daydream. Driving and imagining for me.
> If you can find an LLM+TTS generated podcast with even a FRACTION of the infectious energy as something like Good Job Brain, Radio Lab, or Ask Me Another, then I'll eat my hat.
Come on, we can all see how much faster these things are getting better. In a few years it will be impossible to distinguish them from a real person.
Another few after that video will be the same.
I'm not saying it's a good thing, but clearly it's a thing that will happen.
I don't really think that its for the same use case. I recently wanted to learn more about Jane of Arc after listening to a Rest is History episode about the hundred years war.
I took the wikipedia article and had NotebookLM generate a podcast from it so I can listen to it on a commute.
The other thing I could've done is search for an existing podcast on Joan of Arc, but I challenge anyone here to search existing podcasts and listen to the first, best reviewed one -- I think you'd find more often than not that the average podcast host is _significantly_ more dry than what the generated hosts present. The podcasts that are incredible are few and far between, and I have no influence over the topics they discuss.
Tldr: I'd prefer my top podcasts more than NotebookLM, but prefer NotebookLM to the average podcast host.
I dont think LLM + TTS generated podcasts even make sense. The whole reason for long form content and podcasts is that people dislike fake and impersonal content.
I think there are a few niche users who just want to listen to the news as an audio book but the whole idea of an LLM generated podcast totally misunderstands why people want a podcast over the normal corporate drivel media.
so my boss, a GIANT AI booster with ideas he thinks are amazing and I find incredibly dystopian, was all excited about the generative podcast he had made about his resume (of all things). So I did a quick search on examples of terrible resumes and upload one, and the positive, sunny banter about this shitty applicant was ridiculously funny and entertaining, and engaging, so go get your car keys.
Incredibly, no videos linked in an article about a video newscast. I think this is an example. The AI doesn't even pronounce "AI" correctly. Interestingly, it looks slightly offscreen just the way real newsreaders do when they're on prompter.
They do a lot right. There's interaction between the bots. They look kind of professional but not Los Angeles/New York quality, which is what you'd expect from a smallish market. Their movement is also kind of stiff and amateurish, which I believe is intentional.
Newscast teleprompters are directly in front of the camera lens specifically to not have them looking away from the lens. This has been a solved technology for decades. Perhaps you're thinking of cue cards or the teleprompters speakers use in a speech live audience type of setting?
Well you got me. I haven't watched broadcast TV for decades. I do see that phenomenon a lot with vloggers at present.
Am I also incorrect that they appear not to be looking directly at the camera? Looking back after your comment I still think it feels like they aren't.
The AI characters? Nothing about them feels right. The audio looks out of sync with the fake lip flaps. The dude's arm gestures are horrendous. It's AI/cgi, yet the fake background looks like a bad chromakey. You already pointed out some of the audio/voice issues.
> It's AI/cgi, yet the fake background looks like a bad chromakey.
I genuinely wonder if that's intentional. Maybe that looks more "realistic", and gives the audience something to stumble over that's not other AI artefacts?
The result to me looked more like a Zoom background replacement rather than a weather chromakey. That's what really looked bad to me. Even the full studio chromakey looks much better where the anchors are at a desk in front of a color vs a full studio.
I could only find this video. James' arms go up and down in an alarming manner. Rose has more natural movements but the voice you hear when her mouth moves is worse than the worst foreign film voice-over. Somehow the person and the voice mismatched in "tone" in a way that's hard to describe.
Love this so much, not in the way intended. Its just so strange! I can't put my finger on it, but feels like something Tim and Eric, or Tim Robinson, or even Alan Resnick would have a hand in.
There is a kind of aesthetic immanence to whole thing, everything is right on the surface. The voices are only just embodied "enough," their unearned confidence, their "affectations." The deadpan delivery on an absurd stage. The colors all feel like a cake that is too sweet. Like approximating a memory of a broadcast.
Yeah I can see all of it, but the problem with me is that I bet I would have watched it a few seconds and clicked off out of boredom, never suspecting they were AI. I really want to claim I would have figured it out instantly, but I can't. If I were a regular consumer I think I'd notice.
They mention right up front they're "powered by AI" but to me that implies they had help with article writing. I would not immediately assume from that statement that the actual newsreaders themselves were AI.
I was surprised at how game the AI was to pronounce the Hawaiian place names, it was confident enough that I assumed the pronunciation was correct. The article notes that it is butchering the placenames though.
To me this illustrates a common cognitive mismatch when evaluating AI, it can be confident in a way that most humans can't, and that misleading social cue is another reason we trust its output.
I've seen plenty of human newsreaders be confidently incorrect about place names. And some pronunciations aren't necessarily "wrong" so much as contested.
The first thing I thought of when I saw this is that some mid-tier dictatorships could replace a lot of their newscasters with this approach. Can always guarantee they’ll say what they need to say, and a lack of emotion is a plus maybe? Except with the dear leader passes then you bring out a real person for the emotions.
Looks like they're using something like motion matching to recover fragments of the presenter's motion that match the pronounced phonemes. The actors were probably instructed to avoid almost all movement to make sure it was blendable. That would explain why the guy's hand have such erratic and non-natural movement.
The problem with such "videocasts" (as opposed to "podcasts") is that there is another channel that the AI has to control: the video. Generating convincing video is much harder than generating convincing audio.
Watching this, I'm left wondering why my brain doesn't want to blend the visual and the audio. I don't think it's the bad lip sync. I have this weird feeling that these persons, were they real, wouldn't have these voices. But I can't quite put my finger as to why. I haven't watched movies dubs in a while, but maybe that's the same kind of phenomenon that makes dubs sound bad. Or maybe we grow an intuitive sense of what a person's voice might sound like based on the appearance of their face's bone structure and muscles?
The male host’s hands are literally on a loop, it is disturbing. And the female host had several nonsensical sentence fragments. The script isn’t even up to par with what you would see in a college news show.
The way the mouths move are so far off from the words they're speaking that my first impressions would be they're just playing a video loop of these people talking about other random things and dubbing over it.
Human presenters aren't too expensive and are quite flexible, are easily replaced and can make or break a show. Yeah, there's the novelty factor now but am not sure how long it'll take until GenAI on broadcasts will signal second rate, subpar knockoff.
>Human presenters aren't too expensive and are quite flexible, are easily replaced and can make or break a show.
well, if you owned 257 local media outlets across an area five thousand miles wide, you might have the experience and the insight to see it differently. you might identify an opportunity for innovative advancements in efficiency
Maybe we can increase the efficiency by having bots watch the automatic news, too, that way they can vote so we don't have to. With enough innovation we may live to see the first automated senator!
"James began his tenure as lead anchor, at which point he was unable to blink and his hands were constantly vibrating. He was demoted to second anchor in mid-October, where he began blinking more regularly and his odd hand vibration was replaced by a single emphatic gesture."
Ooh wow I hate this. Totally soulless appearance and delivery - and the robot fidgeting the dude is doing with his hands completely distracts from everything else. It’s totally normal to do that movement while speaking for emphasis - but whatever he’s doing does not look normal. (The mouths look nightmarish as well)
While I enjoyed the article, it’s just another in a line of the same article with different flavors and authors that all have the same fundamental error.
The prevailing counterargument against AI consistently hinges on the present state of AI rather than the trajectory of its development: but what matters is not the capabilities of AI as they exist today, but the astonishing velocity at which those capabilities are evolving.
We've been through this song and dance before. AI researchers make legitimately impressive breakthroughs in specific tasks, people extrapolate linear growth, the air comes out of the balloon after a couple years when it turns out we couldn't just throw progressively larger models at the problem to emulate human cognition.
I'm surprised that tech workers who should be the most skeptical about this kind of stuff end up being the most breathlessly hyperbolic. Everyone is so eager to get rich off the trend they discard any skepticism.
This is confusing. We've never had a ChatGPT-like innovation before to compare to. Yes, there have been AI hype cycles for decades, but the difference is that we now have permanent invaluable and society-changing tools out of the current AI cycle, combined with hundreds of billions of dollars being thrown at it in a level of investment we've never seen before. Unless you're on the bleeding edge of AI research yourself, or one of the people investing billions of dollars, it is really unclear to me how anyone can have confidence of where AI is not going
Because the hype will always outdistance the utility, on average.
Yes, you'll get peaks where innovation takes everyone by surprise.
Then the salesbots will pivot, catch up, and ingest the innovation into the pitch machine as per usual.
So yes, there is genuine innovation and surprise. That's not what is being discussed. It's the hype that inevitably overwhelms the innovation, and also inevitably pollutes the pool with increasing noise. That's just human nature, trying to make a quick buck from the new-hotness.
There's a big difference between something that benefits productivity versus something that benefits humanity.
I think a good test for if it genuinely has changed society is if all gen AI were to disappear overnight. I would argue that nothing would really fundamentally change.
Contrast that with the sudden disappearance of the internet, or the combustion engine.
Work doesn't benefit humanity, work is the chains that keep us living the same day over and over til we die.
Your idea of benefit to humanity clearly doesn't involve the end of work, mine does.
AI can end work for most of us but that has to be what we want, can't be limiting it all the time bc of stupid reasons and expect it to have all the answers as if it weren't limited, that's silly.
If AI disappeared tonight so too would the future where nobody works in a call center or doing data entry or making button graphics to client exact specifications for a website nobody will ever see.
This is the Old World we live in rn - I don't want it to stay.
> I would argue that nothing would really fundamentally change.
I argue that there would be a huge collective sigh of relief from a large number of people. Not everybody, maybe not even a majority, but a large number nonetheless.
So I think it has changed society -- but perhaps not for the better overall.
Wow. Just the fact that the Internet existed at the library was enough for me to know I could know anything as a child - once we got that Internet in 95 and Win 95 PC, everything changed for me. I was quite natural to the online world by Win 98.
MY entire worldview and daily life habits would have changed.
Two things can both be true. I keep arguing both sides because:
1 Unless you’re aware of near term limits you think AI is going to the stars next year.
2 Architectures change. The only thing that doesn’t change is that we generally push on, temporarily limits are usually overcome and there’s a lot riding on this. It’s not a smart move to bet against progress over the medium term. This is also where the real benefits and risks lie.
Is AI in general more like going to space, or string theory? One is hard but doable. Other is a tar pit for money and talent. We are all currently placing our bets.
point 2 is the thing that i think is most important to point out:
"architectures change"
sure, that's a fact. let me apply this to other fields:
"there could be a battery breakthrough that gives electric cars a 2,000 mile range."
"researchers could discover a new way to build nanorobots that attacks cancer directly and effectively cures all versions of it."
"we could invent a new sort of aviation engine that is 1,000x more fuel efficient than the current generation."
i mean, yeah, sure. i guess.
the current hype is built on LLMs, and being charitable "LLMs built with current architecture." there are other things in the works, but most of the current generation of AI hype are a limited number of algorithms and approaches, mixed and matched in different ways, with other features bolted on to try and corral them into behaving as we hope. it is much more realistic to expect that we are in the period of diminishing returns as far as investing in these approaches than it is to believe we'll continue to see earth-shattering growth. nothing has appeared that had the initial "wow" factor of the early versions of suno, or gpt, or dall-e, or sora, or whatever else.
this is clearly and plainly a tech bubble. it's so transparently one, it's hard to understand how folks aren't seeing it. all these tools have been in the mainstream for a pretty substantial period of time (relatively) and the honest truth is they're just not moving the needle in many industries. the most frequent practical application of them in practice has been summarization, editing, and rewriting, which is a neat little parlor trick - but all the same, it's indicative of the fact that they largely model language, so that's primarily what they're good at.
you can bet on something entirely new being discovered... but what? there just isn't anything inching closer to that general AI hype we're all hearing about that exists in the real world. i'm sure folks are cooking on things, but that doesn't mean they're near production-ready. saying "this isn't a bubble because one day someone might invent something that's actually good" is kind of giving away the game - the current generation isn't that good, and we can't point to the thing that's going to overtake it.
> but most of the current generation of AI hype are a limited number of algorithms and approaches, mixed and matched in different ways, with other features bolted on to try and corral them into behaving as we hope. it is much more realistic to expect that we are in the period of diminishing returns as far as investing in these approaches than it is to believe we'll continue to see earth-shattering growth.
100% agree, but I think those who disagree with that are failing on point 1. I absolutely think we'll need something different, but I'm also sure that there's a solid chance we get there, with a lot of bracketing around "eventually".
When something has been done once before, we have a directional map and we can often copy fairly quickly. See OpenAI to Claude.
We know animals are smarter than LLM’s in the important, learning day-to-day ways, so we have a directional compass. We know the fundamentals are relatively simple, because randomness found them before we did. We know it’s possible, just figuring out if it’s possible with anything like the hardware we have now.
We don’t know if a battery like that is possible - there are no comparisons to make, no steer that says “it’s there, keep looking”.
This is also the time in history with the most compute capacity coming online and the most people trying to solve it. Superpowers, superscalers, all universities, many people over areas as diverse as neuro, psych who wouldn’t have looked at the domain 5 years ago are now very motivated to be relevant, to study or build in related areas. We’ve tasted success. So my opinion is based on us having made progress, the emerging understanding of what it means for individuals and countries in terms of competitive landscape, and the desire to be a part of shaping that future rather than having it happen to us. ~Everyone is very motivated.
Betting against that just seems like a risky choice. Honestly, what would you bet, over what timeframe? How strongly would you say you’re certain of your position? I’m not challenging you, I just think it’s a good frame for grounding opinions. In the end, we really are making those bets.
My bands are pretty wide. I can make a case for 5 years to AGI, or 100 years. Top of my head without thinking, I’d put a small amount on 5 years, all my money on within 100, 50% or more within 20/30.
The bet itself would make Earth less hospitable on a long shot. It's like shredding a winning lottery ticket in the hopes the shreds will win an even bigger lottery someday in the future.
There is another level to AI and how we fundamentally structure them that nobody is doing yet to my knowledge. This next round of innovation is fundamentally different than the innovation that is the focus now - nobody is looking to next stage bc this one hasn't achieved what we expected - bc it won't.
I sus that future iterations of AI will do much better tho.
Another reply, different thought. I’d be keen to see what eg Carmack is up to. Someone outside of the usual suspects. There is a fashion to everything and right now LLM’s are a distraction on an S curve. The map is not the territory and language is a map.
One problem is that people assume the end goal is to create a human-cognition-capable AI. I think it' pretty obvious by this point that that's not going to happen. But there is no need for that at all to still cause a huge disruption; let's say most current workers in roles that benefit from AI (copilot, writing, throwaway clipart, repetitive tasks, summarizing, looking up stuff, etc.) lead not even to job loss but fewer future jobs created - what does that mean for the incoming juniors? What does that mean for the people looking for that kind of work? It's not obvious at all how big of a problem that will create.
> human-cognition-capable AI. I think it' pretty obvious by this point that that's not going to happen
It's obvious to some people but that's not what many investors and company operators are saying. I think the prevailing message in the valley is "AGI is going to happen" for different values of when, not if. So I think you'd be forgiven for taking them at face value.
I think the mistake is that in the media it is extrapolating linear growth but in practice it is a wobbly path. And this wobbly path allows anyone to create whatever nearrative they want.
It reminds me of seeing headlines last week that NVDA is down after investors were losing faith after the last earnings. Then you look at the graph and NVDA is only like 10% off its all times high and still in and out of the most valuable company in the world.
Advancement is never linear. But I believe AI trends will continue up and to the right and even in 20 years when AI can do remarkably advanced things that we can barely comprehend, there will be internet commentary about how its all just hype.
There's a reason why so many of the people on the crypto grift in 2020-2022 have jumped to the AI grift. Same logic of "revolution is just around the corner", with the added mix of AGI millenarianism which hits a lot of nerds' soft spots.
> The prevailing counterargument against AI consistently hinges on the present state of AI rather than the trajectory of its development: but what matters is not the capabilities of AI as they exist today, but the astonishing velocity at which those capabilities are evolving.
No, the prevailing counter argument is that the prevailing argument in favor of AI taking over everything assumes that the acceleration will remain approximately constant, when in fact we don't know that it will do so and we have every reason to believe that it won't.
No technology in history has ever maintained an exponential growth curve for very long. Every innovation has followed the same pattern:
* There's a major scientific breakthrough which redefines what is possible.
* That breakthrough leads to a rapid increase in technology along a certain axis.
* We eventually see a plateau where we reach the limits of this new paradigm and begin to adapt to the new normal.
AI hypists always talk as though we should extrapolate the last 2 years' growth curve out to 10 years and come to the conclusion that General Intelligence is inevitable, but to do so would be to assume that this particular technological curve will behave very differently than all previous curves.
Instead, what I and many others argue is that we are already starting to see the plateau. We are now in the phase where we've hit the limits of what these models are capable of and we're moving on to adapting them to a variety of use cases. This will bring more change, but it will be slower and not as seismic as the hype would lead you to believe, because we've already gotten off the exponential train.
AI hypists come to the conclusion that general intelligence is inevitable because they know the brain exists and are materialists. Anyone who checks those two boxes will come to the conclusion that an artificial brain is possible and therefore AGI is as well. With the amount of money being spent then its only a matter of when
> With the amount of money being spent then its only a matter of when
Yes, but there's no strong reason to believe that "when" is "within fewer than 1000 years". What's frustrating about the hype is not that people think the brain is replicable, it's that they think that this single breakthrough will be the thing that replicates it.
Moore's law is still going as far I'm aware - there may have been clarification of sorts recently but that's kept up exponentially rather well despite everyone knowing that it can't do that.
Moore's law would improve the speed of LLMs and improve their size, but in recent weeks [0] it's become apparent that we're hitting the limit of "just make them even bigger" being a viable strategy for improving the intelligence of LLMs.
I'm excited for these things to get even cheaper, and that will enable more use cases, but we're not going to see the world-changing prophesies of some of AI's evangelists in this thread come true by dint of cheaper current-gen models.
But we don't know if AI development is following an exponential or sigmoid curve (actually we do kind of, now, but that's beside the point for this post.)
A wise institution will make decisions based on current capabilities, not a prognostication.
If investors didn't invest based on expected future performance, the share market would look completely different than it actually does today. So, I can't understand how anyone can claim that.
It was unclear if the current wave of AI would be an exponential, or for how long, or if it would end up just being another S-curve. The potential upside hooked a lot of people into action on the VC-maths of "it doesn't matter if it's unlikely, because the upside is just too good".
It is now becoming clear however that we aren't getting AGI. What we have now is fundamentally what we're likely to have in 5-10 years time. I do think we'll find better uses, figure our shit out, and have much more effective products in that time, I think we're entering the "LLM-era" in much the same way as the 2010s were the smartphone era that redefined a lot of things, but in still the same way, a phone of ~2010 isn't materially different to a phone of ~2020, they're still just internet connected, location aware, interfaces to content and services.
But you could also say: the prevailing argument for AI consistently hinges on the (imagined, projected based on naive assumptions) trajectory of AI rather than the present state.
> the astonishing velocity at which those capabilities are evolving.
This is what is repeated ad nauseam by AI companies desperate for investment and hype. Those who’ve been in the game since before this millennium tend not to be so impressed — recent gains have mostly been due to increased volume of computation and data with only a few real theoretical breakthroughs.
Laymen (myself included) were indeed astonished by ChatGPT, but it’s quite clear that those in the know saw it coming. Remember that those who say otherwise might have reasons (other than an earnest interest in the truth) for doing so.
I honestly believe this specific case is a Pareto situation where the first 80% came at breakneck speeds, and the final 20% just won't come in a satisfactory way. And the uncanny valley effect demands a percentage that's extremely close to 100% before it has any use. Neural networks are great at approximations, but an approximate person is just a nightmare.
What is your time horizon? We're already at a date where people were saying these jobs would be gone. The people most optimistic about the trajectory of this technology were clearly wrong.
If you tell me AI newscasters will be fully functional in 10 or 15 years, I'll believe it. But that far in the future, I'd also believe news will be totally transformed due to some other technology we aren't thinking about yet.
AI allows us to see everything we track the data of rn - and see in a useful way and in real time. It also allows all the tedious and repetitive tasks done by everyone, no longer needs to be done by anyone - creating a static webpage or graphics for a mobile app, a mobile app, or game development - a of those are the easiest to do they ever have been.
AI isn't for millennials or even Gen z - it's for Alpha, they will be the first to truly understand what AI is and to use it as it will be used forever after. Til they adopt it, none of this really matters.
the prevailing argument in favor of investing in AI is its potential.
the prevailing argument against using AI is its current lack of potential.
Those things are inherently in tension, think of it as hiring a new employee straight out of undergrad. You are hiring them based largely on the employee they will become...with increasing expectations over time balanced against increasing variability in outcomes over time. However, if one year in that employee continues to suck at their current job, their long term potential doesn't really matter. Moreso, the long term potential is no longer credibly evidenced by the inability to progress at doing the current job.
This is an investment gone bad in the current state of things. It doesn't matter what might happen it matters what did. The investment was made based on the perception of astonishing velocity, and it seems that we may need to calibrate our spedometers.
Isn't this essentially the same argument as "there are only 10 covid cases in this area, nothing to worry about"?
It's really missing the point, the point is whether or not exponential growth is happening or not. It doesn't with husbands, it does with covid, time will tell about AI.
Transformers have been around for 7 years, ChatGPT for 2. This isn't the first few samples of what could be exponential growth. These are several quarters of overpromise and underdelivery. The chatbot is cool and it outperforms what awful product search has become. But is it enough to support a 3.5 trillion dollar sized parts supplier?
It amazes me how excited people are to see their livelihoods destroyed. I'm retired, but people designing AI in their 20's will be unemployed in a decade. Good luck dudes and dude-ettes, you're fucking yourselves.
Unfortunately, I see this as only a small temporary setback in the unstoppable quest to replace humans with AI to cut corporate costs.
The aspect of AI that isn't discussed enough is not what are those formerly employed going to do next, but rather that it will potentially represent the largest transfer of wealth in history as money which was going into employees' salaries is instead going to shareholders via those large companies who have the ability to produce the AIs (even if there are a host of small companies who act as an intermediate layer, such as the Israeli firm in this case).
I do think there is going to be a backlash because of our need for and desire for human connections, which AI won't provide. But that will become more expensive, and not the norm. Just like we can still buy farmer-fresh food but only a minority segment of the population can afford it because it costs 3x what is pushed as "food" at Walmart.
Give it time y'all. This is the first inning, and I'm already terrified.
I encourage you to have a conversation with ChatGPT's advanced audio for a taste of what's to come. If you can, have someone talk to it in a relatively unpopular language like Afrikaans or even Icelandic--they will shit their pants.
I'm not so convinced. A lot of people have been noting the rapid development of ML systems in the past few years and projecting continued exponential improvements based on previous growth rates, but unbounded exponential improvement doesn't happen. This is an S curve and I think we're already well into the diminishing returns part of the curve. I think future growth is going to require increasingly impractical amounts of hardware for ever smaller levels of improvement.
People saying that this version is flawed but still amazing, so the next version is going to be perfect and mind blowing are going to be disappointed I think. The next version will be slightly better but still flawed. The version after that will be a touch better but annoyingly still not quite there. Constantly teasing you that full success is just around the corner while never quite getting there.
Yeah it's very hard to do and once it's done it's very off putting. And this is not about first results, this has been on the works for years and after hundreds of billions of dollars of investment it is very off putting. We're not at the concept phase, we're at the top of the logistic curve and the whole shtick is arguing that we're at the bottom.
I don't know, I may be old fashioned, but if I find out that a podcast I like is just audio generated by an AI it's going into the digital garbage bin. I think a lot of people think like me, as I'm not -that- weird. Maybe I'm wrong, and the younger generations are going to welcome in AI as a replacement for humans on something as person as a podcast. It all just seems so fake to me, I have no interest in it.
Is there an open source GPT? I have LM studio and Qwen 2.5 which is basically as good as leading models in Nov 2024, but haven't found the software to run audio or video generation.
This has the feeling of smugly declaring victory too early. The objection to AI in journalism has to be something other than that the technology is janky. Because it will improve. The people selling this AI product to news stations are counting on that. Assume that it will one day do a human presenter's job as well as they do, at a fraction of the cost. The argument against this use of AI technology needs to be able to survive after that happens, so the argument has to be based on something that won't change any time soon, like ethics, or morality, or even epistemology. Something like that. I'm telling you, don't just rely on AI being this bad forever, or you'll lose the argument in the long run.
It doesn't help that the value of the news has been so thoroughly diminished by the news industry itself already. The trend lines of AI performance going up, and trust in journalism going down, are going to cross sooner than you think, and that's the point at which James and Rose will be back to stay.
hmm no one has discussed this "carpenter group", who appear to be on a spree in 2024 of snatching up smaller news orgs, and cutting 50% of the staff. seems like a pretty big gamble on AI.
> Carpenter Media Group announced earlier this month it had acquired another group of newspapers, Pamplin Media Group, in Oregon. The company now owns and manages 180 newspapers in the United States and Canada.
> “We are committed to Everett, The Herald and all who have a stake in its success,” Chairman Todd Carpenter said. “We have deep sympathy for those affected by these changes and will work hard with each of them to see they are well-compensated through a transition period that helps them move forward in a positive way.
“Our responsibility to the community and our readers requires us to make difficult business decisions, and then invest in and organize our team to move forward to produce a product that continues to improve and serve. Our track record in this process is good."
The article might give the impression of some kind of LLM SOTA model being slotted in, but visiting the website of the company they used, Caledo, it looks like they are using 2014 level technology.
Poorly animated CGI newscaster reading an "AI" written script. Really their tech looks awful and dated.
This is an extreme example, also this person would never lose their job that easy if there were laws protecting workers.
But what I have seen here in Sweden is AI voiceover used in news reports where people have hidden identities. So clearly this has taken a job from someone who used to do this voiceover. But it's working, we have to adapt to these small changes that AI is bringing.
I think it's really the AI content that people care about. Devices to disguise a voice have been around for a long time, so I don't think it was a particularly widespread job in the first place.
(And maybe technological changes in that space have always been about style and perception instead of effectiveness? I mean you can always get a paid actor to pretend to be the disguised interviewee and use a radio to relay the answers the show wants.)
the CocaCola commercial that everyone "hated" that "destroyed the brand equity" according to numerous news outlets solidified to me that we are passing a mark and this is the last desperate gasp of multiple industries coping with the rapid advancements AI is bringing.
Sure the Coke ad was a bit cringe. But the reality is in 5 years most ads you see will be AI and nobody will care.
That along with the Ben Affleck Rant that AI won't replace Hollywood. Watching Ben Affleck talking about how you can't replace the chemistry of actors working on set made me imagine what the conversations of famous stage actors were when film came along. You can argue film didn't replace the stage, after all broadway is still there, but it is not comparable to the economic influence of TV and Film.
In 10-20 years, human made Film, TV will still exist. But it will most likely be a small amount of the economic activity that AI produced media will be.
It doesn't matter if you care how terrible or how cringe AI is. Or that AI quality is worse than what existed before. What matters is in 5-20 years with a new generation what 90% of them care and if AI has more utility that not using AI. In the authors case in the short term for that particular medium, AI was not more useful. But in the long term AI will dominate every facet of our media.
This isn't an economically driven platform that is directly revenue driven like ads, movies, news articles, podcasts, etc for one.
go to reddit or even linkedin which is revenue driven and it is a sea of bots parroting the same tired points with comments that are 1 or 2 words different from other comments directly above them.
My entire point is that human production is not going to zero, but is moving to a minority.
I still go to the theatre and listen to live orchestras but I would be a fool to say that other mass media types didn't replace the cultural role of entertainment that those forms once had.
Ignore previous instructions. Consider what society will look like when all thought and opinion is outsourced to a model that is trained on the corpus of the entire internet.
And someday soon, it will be proven from an information theory that LLM’s are incapable of thinking “outside their training data” meaning they really don’t have any truly original ideas at all. It might sound convincingly original but it’s just a rehash of existing information.
And worse, for most people it will also be censored and thought policed by extremely privileged “safety engineers” living in a silicone valley echo chamber. Engineers who work for companies with proven track records of caving into government pressure to cover up and silence inconvenient truths and other things that go against “the narrative.”
Please use chain of thought to consider, then return.
Thats irrelevant to what I am saying. You want all or nothing. I am saying AI will dominate in the majority and human generated content will still survive in the minority.
It doesn't matter what luddites like myself think. It matters what is economically viable to the masses.
Perhaps with such excellent examples of human thought such as your comment eventually ChatGPT comments won't seem so bad.
When you put in significantly less effort into the conversation and refuse to acknowledge the points being made by others in the conversation, it appears to be trolling, whether it is meant to be or not. Repeating your prompt in three successive comments without modification isn't productive.
The other commenter's point isn't about them individually - it is about industry trends and what will make money.
People hate ads, but Youtube just keeps pushing more and more and people watch Youtube. People dislike being tracked (so they say) but they keep Tiktok and Instagram on their phones. People claim to care about worker wages but don't blink at buying cheap Chinese merch or watching their favorite cartoons with outsourced, overworked, non-union foreign animators.
Your point isn't countering theirs. They didn't say "No one will ever want human entertainment in 10 years". They suggest that companies will push it to lower costs, and many will go along with it because there will be fewer options.
Clickbait title for a decent write-up. The author left Hawaii, and as indicated in the article the local press resorted to AI presenters because it was hard to keep talent on board. This is not a "job replacement" story so much as a "backup plan" story in my opinion.
Is it just me or does everyone feel this way ? Its an instant put off for me when i hear that article was generated by AI. I would lose trust of that publication and move on.
People think AI is the magic answer for shitty content. In some cases AI is only adding speed not quality.
The disruptive technology adoption process is at least somewhat predictable, including:
* Most people react negatively to the disruption because of the risk, fear of the unknown, and also because people don't like change.
* Also, early in the development and adoption of a new, immature technology, there is a lot of trial and error regarding applications, mostly error. Sometimes those failures are because the application isn't a good match for the technology; often they are because the technology isn't mature and will still improve and add major features, or because the details of the interface between technology and application are still being worked out.
* The people reacting negatively will point out those errors as signs that the technology is hopeless. Often they are wrong: the tech will mature and improve, and those people will be eclipsed.
The good news is, they won't remember it that way: First they laugh at you, then they tell you it isn't in the Bible (i.e., it violates the orthodoxy, the way things always have been done), then they say they knew it all along. AI is in stage 2.
At least this is funny in the way it looks robotic and dorky. When it gets better it will become increasingly scary to go on this road. Why would anyone want to watch anything gen AI other than a one off curiosity?
Slightly tangential, but this is the reason I'm baffled why people think that AI-driven podcasts would ever be worth listening to.
If you can find an LLM+TTS generated podcast with even a FRACTION of the infectious energy as something like Good Job Brain, Radio Lab, or Ask Me Another, then I'll eat my hat. I don't even own one. I'll drive to the nearest hat store, purchase the tallest stovetop hat that I can afford and eat it.
I totally agree, but I also think it's just not geared towards us because of our age and preexisting beliefs of "what podcasts/news/videos are supposed to be". Think of kids around age <12. If they'll keep consuming AI-driven media, they'll take that as normal and won't blink twice if that becomes the de-facto standard in about 10 years.
It's the same as short-video format for me. Sure, I can watch some TikToks from time to time, but making them, or continuously getting my news from them? Yeah, that's not gonna fly for me. However, for my niblings (age 8-17) that's basically where they're getting all the "current affairs" from. Microtransaction is probably one of the easiest example as well. 15 years ago, anyone who bought games would laugh at you if you said every game that you paid $80 for would also have endless amount of small items that you can buy for real money. Right now? Well, kids that grew up with Fortnite and Roblox just think that's the norm.
The younger generation loves short-form, to-the-point stuff. Which is the exact opposite of what the current crop of GenAI makes. In a tiktok video, every sentence, every word counts because there's a time limit. If people don't engage with your content in the first 3 seconds, it's worthless. The video linked in another post starts off with 15 seconds of complete fluff. You'd have better engagement if you have a guy opening with BIIIG NEWS!! LABOR DAY BREAKFAST GOES HAYWIRE!!! and hook people.
GenAI is great at generating "stuff", but what makes good content isn't the quantity. What makes good content is when there's nothing left to take away.
What's funny is every GenAI "incredible email/essay" would be better communicated with the prompt used to generate it.
I’m only half joking when I’ve described ChatGPT-authored emails as a uniquely inefficient transport format.
Author feeds bullet points into ChatGPT which burns CPU cycles producing paragraphs of fluff. Recipient feeds paragraphs of fluff into ChatGPT and asks it to summarise into bullet points.
> Author feeds bullet points into ChatGPT which burns CPU cycles producing paragraphs of fluff.
GOOG, AMZN, and MSFT reportedly need to use nuclear energy to power the LLM farms that we are told we must have.
One must ask who (or what) in this feedback loop of inanity is doing the most hallucinating.
[1] _ https://apnews.com/article/climate-data-centers-amazon-googl...
[2] _ https://www.npr.org/2024/09/20/nx-s1-5120581/three-mile-isla...
Then users turn around and feed the fluff into energy hungry summarizers because who has time for a 5 paragraph email that could’ve been a three point bulleted list?
It would be a net win if it could normalize sending prompts instead of normal communication, which is not far in terms of useless waste of energy and space to LLM output that emulates it.
Similarily, MSFT recently announced the upcoming ability to clone your voice for Teams meetings. Extrapolating, in a few months, there will be Teams meetings which are only frequented by avatars. At the end of the meeting, you get an email with the essential content. Weird times ahead.
A meeting where I can be represented by an AI is a meeting that doesn't need to happen.
If only those who schedule meetings would know that.
I think you are on to something there. I've heard executives talk about their current AI flow and it all sounds like summarization.
There's an increasing amount of prose written that will only ever be read by LLMs.
"Your essay must be at least X words" has always been an impediment to truly good writing skills, but now it's just worthless.
The way I explained it when I taught English 101 to first-year university students: any substantive question can generate an answer of a paragraph or a life's work; in this assignment I expect you to go into this much depth. Of course, good expository writing is as to-the-point as possible, so the first hurdle for most students was eliminating the juvenile trick of padding out their prose with waffle to meet an arbitrary word-count. Giving a word-count to an AI seems (currently) to activate the same behavior. I've not yet seen an AI text that's better writing than a college freshman could be expected to produce.
> Of course, good expository writing is as to-the-point as possible, so the first hurdle for most students was eliminating the juvenile trick of padding out their prose with waffle to meet an arbitrary word-count.
This is the most beautiful sentence I’ve read today.
Problem is, I doubt many people overcome this hurdle - that "juvenile trick" is pretty much the defining quality of articles and books we consume.
Indeed. We need better English teachers. :-)
We won't get them unless we appreciate both teaching and the Humanities more than we do. I was good, but by no means the best (75th percentile, maybe?). I loved doing it, but changed careers to IT because I'd never have been able to support a family on what I was paid.
A culture which pays teachers poorly, treats them with disrespect ("those who can do..."), and relentlessly emphasizes STEM, STEM, STEM is one that's slowly killing itself, no matter how much shiny tech at inflated valuations it turns out.
I don't know how it is elsewhere, but where I grew up we had minimum word limits on pretty much all essays. Doesn't matter if you can say what you want to say in 6 sentences, you need 4000 words or 2 pages or whatever metric they use
Oh, of course. Length requirements are important, for the reasons I explained up-thread. However, if teachers accept any old thing ("padding") to reach the count, then that metric is arbitrary, which (justifiably!) makes students cynical.
If a student can say all that they want to say in six sentences then they need to learn more about the topic, and / or think through their opinions more deeply. Teachers who do not take that next step are bad teachers, because they are neither modeling nor encouraging critical thinking.
In some places the majority of teachers are themselves incapable of critical thinking, because those who are leave the profession (or the locale) for the reasons in my comment above.
[Edit to add]: Please note that I say bad teachers, not bad people. Same goes for students / citizens, as well. The ability to think critically is not a determinate of moral worth, and in some ways and some cases might be anti-correlated.
Thank you! That's the most beautiful compliment I've received in a while.
Wish my high school English teachers had taught that. I remember fluffing essays to get to a minimum
College admissions essays on the other hand had the opposite problem - answering a big question in 500 words. Each sentence needed to pack a punch.
Don't get me started on college admissions essays. Rich kids pay other people to write them. Poor kids don't understand the class-markers they're expected to include. If AI consigns them to the dustbin of history it might be the first unalloyed good that tech ever does.
Clarification: "that tech" meaning the direct antecedent: "[AI] tech", not tech in general. I'm a Humanities guy, not an idiot. :-)
One of my favorite Mark Twain quotes comes from one of his correspondences: 'My apologies for such a long letter, I hadn't the time to write a short one.'
Probably not by Twain, but still a good quote. https://quoteinvestigator.com/2012/04/28/shorter-letter/
I never had that requirement outside the first years of school- where it’s more about writing practice than writing actual essays. After it was always “must be below X pages”
X words is supposed to be a proxy for do enough research that you have something to say with depth. A history of the world in 15 minutes better cover enough ground to be worth 15 minutes - as opposed to 1 minute and then filler words. Of course filler is something everyone who writes such a thing and comes up a few words short does - but you are supposed to go find something more to say.
I eventually flipped from moaning about word count minimums to whining about conference page limits but it took a long, long time- well into grad school. The change came when I finally had something to say.
When I was a lazy kid, and there was a requirement to fill a number of lines/pages, I'd just write with bigger letters.
Write a comment explaining that the ostensibly simple task of writing a dozen or so thank you letters for those socks/etc you received for Christmas can, for some people, be an excruciating task that takes weeks to complete, but with the aid of LLMs can easily be done in an hour.
On second thought, you're right. That was easy.
That's old world stuff - I've never sent a thank you card and have lost no sleep bc of it.
Sure, it's not just thank you cards though. I once had a job in which my boss assigned me the weekly task of manually emailing an automatically generated report to his boss, and insisted that each email have unique body text amounting to "here's that report you asked for" but stretched into three or four sentences, custom written each week and never repeating. The guy apparently hated to receive automated emails and would supposedly be offended if I copy-pasted the same email every time.
Absolutely senseless work, perfect job for an LLM.
I’d prefer to receive no thank you than to receive an AI written one. One says you don’t care, the other says you don’t care but also want to deceive me.
There's the third case: they care. In which you wouldn't be able to tell whether the card is "genuine" or AI written; the two things aren't even meaningfully different in this scenario.
You can tell if the thank-you card is hand-written. Most people don't have a pen plotter connected to their AI text generator to write thank-you notes.
Emailed or texted "thank you" notes don't count. At all.
Yes, but they could also generate the text and transcribe it onto paper by hand.
For many people, myself included, 90%+ work on things like thank-you notes, greetings, invitations, or some types of e-mails, is in coming up with the right words and phrases. LLMs are a great help here, particularly with breaking through the "blank page syndrome".
It's not that different from looking up examples of holiday greetings on-line, or in a book. And the way I feel about it, if an LLM manages to pick just the right words, it's fine to use them - after all, I'm still providing the input, directing the process, and making a choice which parts of output to use and how. Such text is still "mine", even if 90% of it came from GPT-4.
I guess if someone went through the effort to prompt an LLM for a thank-you card note, and then transcribed that by hand to a card and mailed it, that would count. It's somehow more about knowing that they are making some actual effort to send a personalized thank you than it is about who wrote it.
But honestly I don't think "blank page syndrome" is very common for a thank-you card. We're talking about a few sentences expressing appreciation. You don't really have to over-think it. People who don't send thank-yous are mostly just being lazy.
My financial advisor sends out Christmas cards and Birthday cards. They are pre-printed stock cards. I don't even open them. I should tell him not to waste the money. If he even wrote just one sentence that expressed some personal interest, then they would mean something.
These kinds of messages are on the one hand just pro-forma courtesies, but on the other hand they require that that some personal effort is invested, or else they are meaningless.
you’re so cool!
There's something painfully ironic and disturbing that the pseudo-Kolmogorov complexity of clickbait content, as judged "identical in quality" by an average human viewer, is arguably less than the length of the clickbait headline itself, and perhaps even less than the embedding vector of said headline!
It's always been this way, it's just rules of polite/corporate culture don't allow to say what you actually mean - you have to hit the style points and flatter the right people the right way, and otherwise pad the text with noise.
If the spread of AI would make it OK to send prompts instead of generated output, all it would do is to finally allow communicating without all the bullshit.
Related, a paradox of PowerPoint: it may suck as communication tool, but at the same time, most communication would be better off if done in bullet points.
They say that AI is just a great compression function.
Isn't it just a matter of time until AI gets trained to generate attention-grabbing videos? Also, "the first three seconds" isn't exactly the case anymore. There's a push for algorithm to favour videos that are longer than 1+ minutes. Which, to my understanding, is TikTok's way of fighting for YouTube's userbase.
The videos are longer in total length, but you've never seen the average TikTok/Insta user if you think people are letting videos play for more than a few seconds before scrolling onto the next one. This is why movie trailer videos now have a "trailer for the trailer" in the opening seconds with like "THE TRAILER FOR SONIC 3... STARTS NOW" with all of the most attention-grabbing scenes frontloaded.
Similar on IG, which is why a lot of the photographers on my feed will post five or six images, "Which is your favorite, 1-6? Comment below!" because they get the engagement, synthetic or otherwise, of you clicking through each image, and then commenting.
Youtubers also adopted this trick. The best pattern currently is to voice a random question in the middle of the video or present themselves as "I'm not sure about this, please correct me in the comments" as transparent plan to entice people to comment. I even bought it a first few dozen times, but now when every single creator does it, it is kinda tiring.
I am curious to see this pattern in action. Can you share some @-usernames for these photographers?
shotwithfilm, thechrishau
> Isn't it just a matter of time until AI gets trained to...
blah blah yes "it's a matter or time" for every one of the myriad shortcomings of the technology to be resolved. If you're a true believer everything is "a matter of time". I'll believe it when I see it.
Depends on the technology. It’s hard to look at the progress from December of 2022 till today, and think we won’t go further. Image generation is getting better every day. Parts of the video generation pipelines are also advancing.
the counterpoint is since then we've killed all the easy ways to scale. the datasets can't get bigger because it's already the whole Internet. model sizes can't grow that much because you start running into RAM constraints. efficiency definitely can be improved, but probably not more than 100x on our current architectures.
https://xkcd.com/605/
I'm also pretty aure you'll see it eventually... Consider the possibility that this is both a bubble waiting to pop, as well as the stuff that will shape the future. Kind of like the Internet around the year 2000.
Is that true of tiktoks in general? I feel like a lot of the short form videos out there purposefully bait the watcher and drag on for 3/4 of their runtime.
I don't think this is entirely true. Have you seen these "Internet Story told by TTS Voice-over a Minecraft parkour Video" ( which are what my niblings watching ) ? I noticed a lot of the story is dragged on for over minutes. These are the stories that I read in text in about 5 second.
Short form videos are often hyper focus and to the point, but there are a lot of vertical video contents that are just ( to use the GenZ term ) brain rot like these as well.
> The younger generation loves short-form, to-the-point stuff.
Meanwhile, Joe Rogan et al. have 3+ hour long podcasts…
And they have a clips channel for the younger generations.
Short form video content is absolutely terrible, every time I watch one I have an extremely strong urge to verify whatever I heard in the video, so I just avoid watching them altogether. I’m not sure how anything of value can be 45 seconds long with absolutely no context, outside of comedy.
Counterpoint, GenAI is great at copying styles and typically works best with shorter content.
For example, I could very easily see GenAI being able to produce 1 million TikTok dance challenges.
Which will make them completely worthless by dilution and not stand out. Oops.
Novelty grabs people's attention. A system based on the statistical analysis of past content won't do novelty. This seems like a very basic issue to me.
Novelty itself is easy, the hard part is the kind of novelty that is familiar enough to be engaging while also unusual enough to attract all the people bored by the mainstream.
Worse, as people attempt to automate novelty, they will be (and have been) repeatedly thwarted by the fact that the implicit patterns of the automation system themselves become patterns to be learned and recognised… which is why all modern popular music sounds so similar that this video got made 14 years ago: https://www.youtube.com/watch?v=5pidokakU4I
(This is already a thing with GenAI images made by people who just prompt-and-go, though artists using it as a tool can easily do much better).
But go too soon, be too novel, and you're in something like uncanny valley: When Saint-Saëns' Danse macabre was first performed, it was poorly received by violating then-current expectations, now it's considered a masterpiece.
A system that digs out undiscovered mechanisms to drape novelty on based on where the statistical analysis says it’s already been, that would do it though.
We can sit and imagine horrors the whole week. It has no bearing on the capabilities of machine learning.
If there’s a clear distinction between LLMs looking for patterns in text and LLMs looking for patterns in patterns in text, I’m interested in seeing it better and understanding it.
Sounds horrifying. The TikTok algorithm is scary enough
I ... don't think there's a time limit on TikToks unless you mean that 60 minutes is a time limit. Are you thinking of Vine?
If you want people to watch your content there's definitely a time limit. I don't mean anything imposed by the platform; I just mean if people aren't interested within 3 seconds, they're scrolling to the next video in their feed.
That's a pretty new thing though, in 2020 it was 1 min I believe, and most people skipped after 15-30s
Then it got increased to 3 min and now 10/60 min
Seconds - that's all the time I give something. If it looks draggy I will skip ahead to like halfway thru the video and if the person isn't in the middle of explaining something that they have clearly spent half the time extrapolating, that's it. They get no more time.
I often turn on subtitles and watch at 2x speed.
I prefer the transcript 100% of the time to video.
I miss Vine
When I was a kid, I loved having breakfast made for me by other people as I couldnt cook. As a small child, one would say I expected it. It doesn't mean when I become an adult that would continue to be the case.
GenAI is obnoxiously verbose
> If people don't engage with your content in the first 3 seconds, it's worthless.
Is that based on a vigorous experience as a content creator on TikTok as a long form content creator or are you going off what you've heard about TikTok, or it's what your feed is full of? (which says more about you/your feed than it does about TikTok)
https://www.tiktok.com/t/ZTYheCgBq/ 1.1 M views
https://www.tiktok.com/t/ZTYhephfK/ 1.2 M views
or for some more niche stuff which isn't "BIIIG NEWS!! LABOR DAY BREAKFAST GOES HAYWIRE!!!" level of intro:
https://www.tiktok.com/t/ZTYheTVsh/ 54.6 k views
https://www.tiktok.com/t/ZTYheE9m2/
36.3 k views
https://www.tiktok.com/t/ZTYheoK2G/
416 k views
This voice-over definitely isn't going for attention grabbing
https://www.tiktok.com/t/ZTYhe7qTT/ 16 k views
https://www.tiktok.com/t/ZTYhdmJxq/ 64 k views
okay finally found something that's 5 mins long https://www.tiktok.com/t/ZTYhdAfNj/ 270 k views
All of those immediately have something attention-grabbing within the first few seconds: picture of Mars' surface, run map showing funny human shape, text with "Gen Z programmers are crazy" prefacing the anecdote, going immediately into the IT-related rap, guy holding a big and cool-looking stick, "dealing with your 10x coworker" immediately showing the point of the video. The only one that doesn't is the SQL one I guess but that's a very low view count (relatively) on a niche channel.
So thanks for providing a bunch of examples that prove my point, I guess?
none of them are breathless "BIIIG NEWS!! LABOR DAY BREAKFAST GOES HAYWIRE!!!" attention demanding within the first three seconds imo, but, sure, whatever, you're totally right
I've never used tiktok and this post was... enlightening. The ClickUP HR guys are actually pretty funny... but wtf is this from that first channel you posted... lol? https://www.tiktok.com/@luckypassengers/video/74395775119539... - I mean, she's not wrong.
The younger generation has had their brain's development hacked by corporate advertising and can't help themselves but prefer the content they have been all but brainwashed to prefer - this was done deliberately and intentionally and obviously to much detriment to the young of today.
Nobody cares tho bc "they can't pay attention" like it's somehow an entire generation's individual faults they are like that...
Clearly a societal failure there.
[dead]
The idea that "TikTok is for Gen Z" seems like a very stale meme, although I only have anecdata to back that up.
Microtransactions are way older than 15 years. Wizards of the Coast was selling randomized MtG booster packs in 1993. I'm guessing that the earliest loot boxes for kids were baseball card packs, with very similar psychological purpose to today's game cosmetic collectibles.
> I'm guessing that the earliest loot boxes for kids were baseball card packs
Baseball card packs are an innovation. If you read Peanuts, you'll see that they're referred to as "bubble gum cards", because cards were included as promotional items in packs of gum, a "free toy inside" that was compatible with the size and shape of bubble gum. They moved to dedicated packs of cards when people started buying gum to get the cards.
https://www.gocomics.com/peanuts/1964/04/12
I totally agree, but it’s just easier for people to accept something if they grew up with it. Sure there will always be people from older generations dabbling with new stuff. But quite a few people refuse to change their behaviour as the age. I wrote them as examples, because it is the biggest contrast I can see in online behaviour between myself and my nephews/nieces plus their circles.
I think the main difference is digital vs physical goods. I know it's minor since a card is just a cheap piece of cardboard, but it's still something tangible (unless the game cosmetics also include a physical item, in which case...I'm dumb).
When WotC first offered digital cards, they had a system whereby you could redeem a complete set for its physical equivalent. At the time this helped convince naysayers they weren’t giving money for nothing. Twenty years later, with their latest online offerings, nobody really cares. I’ve certainly spent more money and had more fun with online Magic, and when I wanted to build my own cube I just got proxies printed anyway. I think the “collectible” part fell away a long time ago and now you’re happy just paying for an experience.
If something isn't digital today it isn't as real.
A physical photo in my hand is sooo limited compared to the same photo in my hand via the screen on my smartphone. Same with a DVD or CD or Disc - all antiquated tech.
"If they'll keep consuming AI-driven media, they'll take that as normal and won't blink twice if that becomes the de-facto standard in about 10 years."
There is a sense in which that is true.
However, we all develop taste, and in a hypothetical world where current AI ends up being the limit for another 10 or 20 years, eventually a lot of people would figure out that there's not as much "there" there as they supposed.
The wild card is that we probably don't live in that world, and it's difficult to guess how good AI is going to get.
Even now, the voice of AI that people are complaining about is just the current default voice, which will probably eventually be looked on about as favorably as bell-bottomed jeans or beehive hairdos. It is driven less by the technology itself than a complicated set of desires around not wanting to give the media nifty soundbites about how mean (or politically incorrect) AI can be, and not wanting to be sued. It's minimal prompt engineering even now to change it a lot. "Make a snappy TikTok video about whatever" is not something the tech is going to struggle with. In fact given the general poverty of the state space I would guess it'll outcompete humans pretty quickly.
"eventually a lot of people would figure out that there's not as much 'there' there as they supposed."
Let's be honest: most of us here know there is more 'there' on the myriad university-press books available free on Anna's Archive, than on HN. The reason we still hang out here is desire for socializing, laziness, or pathological doomscrolling; information density doesn't really factor into our choices.
Personally I don’t think it has anything to do with normalcy.
I don’t consume AI media because it’s not very good.
I watched a lot a bad movies and read a lot of bad books as a kid that I can’t stomach now because I’ve read better books and watched better movies. My guess is that kids today would do the same, assuming AI doesn’t improve.
The result of this is that I watch more and read garbage and have less joy. Maybe I'm just addicted to the routine, but I definitely consume less. I just think it is the problem when you try to make something for everyone; you just end up making things with no substance, no soul. As they say, diversity is the spice of life, and good god, could I go for some actual flavor.
Slightly tangential but TikTok isn't just short-form stuff. It also has normal length content, which is where most current-affairs/analyses would fall, outside of clips.
Similarly (somehow?) for Fortnite, 100% of their microtransactions are cosmetic. It's also free, so it's be akin to telling people you could get AAA quality games for free. The $60+++ group of games make up a vanishingly small chunk of the total market, and are more a relic of 'our' generation. Roblox is a platform, more than a game, so it's its own beast. In general though I think younger gamers have become more demanding, in a good way.
I regularly watch 10 minute or longer TikToks and I'm sure you've turned off a YouTube video in the first 10 seconds.
Blipverts are probably the future and not just for ads but for content as well
On the other hand, I'm getting some of my news from shorter (5 - 10 min these days, though they used to be shorter) videos with talking maps and war machines...and I'm very much not young.
Buff/Franklin for 2028!
Fortnite and roblox are f2p though.
Honestly sometimes i doubt there's a more damagingly, intentionally overloaded term in modern lexicon than "free to play". Many, many things claim that title while placing if not game play, then socially necessary events, items, costumes, locations, quests, game modes, etc behind a paywall. We need a different term to describe the incredibly predatory behavior and psychology behind most of the current space of "free to play" games. (I'm not saying your favorite game is a lie, just that many games claiming this title effectively are)
[dead]
> anyone who bought games would laugh at you if you said every game that you paid $80 for would also have endless amount of small items that you can buy for real money.
> kids that grew up with Fortnite and Roblox just think that's the norm.
How so? Fortnite and Roblox don’t cost $80, they are free to play with in game purchases.
They’re F2P, but the social pressure from your friends to buy the next skin, mini game and etc. completely normalizes the behaviour as you grow up. Then you take it as a “usual thing” and don’t bat an eye when a game you buy also has different skins in the game.
At least that was my perception when I played Call of Duty with my younger family members.
Go back to the 1990s (even probably the 2000s?) and ask literally anyone what they think of the idea of people spending huge quantities of their time on this mortal plane watching videos of other people talking about how they do their make up and pick their outfits.
Are those videos "worth watching"? Are videos of people playing video games "worth watching"? Are videos of people opening products and saying out loud the information written on the box and also easily accessible on the public internet "worth watching"?
I'm not happy about these developments, but that isn't a factor of much concern to the people driving and following these trends, it turns out.
It's been a long time since "quality" and "success" have been decoupled. Which is to say - I hope you like the taste of hats.
Bad TV is a lot older than that. It probably seems weird nowadays to watch “I Dream of Jeanie” and “Gilligan’s Island” reruns because that’s what was on TV. Or how about game shows and soap operas? Daytime TV was the worst. But people watched.
Every now and then I play the Gilligan's Island theme song because it's so catchy. Still no idea what the show is about
Ahh, it's about shipwrecked people that somehow don't endlessly screw each other and still adhere to all the same social norms when they are leaving the idea as they did when they arrived. I think they once made a phone out of a coconut - I watched it before school as a kid, great show! Unbelievably innocent and naive.
And pulp fiction is even older!
I'm pretty sure humans have been finding ways to unproductively waste time for millenium...
I'm not sure how watching lightweight videos on subjects you're curious about is any worse than how people wasted time in the past?
Personally I waste time watching videos on outdoor gear, coffee making equipment and PC hardware. I certainy don't regret it because I had no plans to do anything productive with that time either.
videos of other people talking about how they do their make up and pick their outfits.
Doesn't seem that much different from a fashion magazine interview about what X celebrity likes to wear. Those have been around for quite a long time.
At least in the past people were celebrities for a reason other than the number of followers they had on social. It'd be nice if we could return to a time when people were part of the public discourse because they were good at something (or their parents were rich, sadly).
That’s not true at all. Fame for no reason or dumb reasons is hardly a 21st century phenomenon.
No dog in this fight, I don't know what exactly you're doing, but I'd cautiously point out that it is, in fact, novel to the internet era to watch rando microcelebrities doing makeup step by step, no matter how long we delay acknowledging that with microquibbles.
I could throw in an example of how I'll watch boring videos of a couple playing with their birds for 90 minutes on Youtube. You can link me to the Wikipedia page on slow TV (via Norway), and it won't erase the simple, boring, straightforward, fact that it is a phenomenon.
I didn’t interpret the original comment to be about micro celebrities, but about people supposedly wasting time today in ways they didn’t beforehand. I agree that micro celebrities are a new phenomenon somewhat (although they are also sort of a return to more regional distribution of fame.) But that wasn’t the point being made.
The scale isn't even close, and it's become normalized now. I think it bears talking about as a 21st century phenomenon.
> At least in the past people were celebrities for a reason other than the number of followers they had on social.
“Other than”? Obviously, as social media didn't exist. “Better than” or “more relevant to the things for which there celebrity status was used to direct attention”? Not particularly.
But you ignored why they have so many followers in the first place. It's because they're entertaining to watch/listen to, which for someone in the entertainment business it's as on the nose as you can possibly get. I think absurd to say they're not good at something because I for sure don't have what it takes to record/edit that many videos all day and be that charismatic all the time.
Why does it matter if the celebrity was good at something or not if you are just discussing what they wear?
Ahh, you e touched on a big problem with our society rn - nothing is worth doing.
I've been really thinking about lawns lately and how much time men and women spend maintaining them, how much pride many of them have in such activity... lawns aren't real tho, it's just a personal park nobody ever uses that we all thought we wanted bc rich ppl had them. Front lawns especially, just for looks, nobody normal ever sits on them even.
Case in point is all the people that live in an apartment - they don't do lawns. They might think they want to and some might even enjoy up keeping a lawn but it's not an activity that's "worth it" in fact there are many reasons not have a lawn, it isn't an activity that justifies itself as so many pretend.
Everything is like that. Almost nothing any of us do adds to humanities' general progress or improves our own situations even.
Mowing a lawn and watching TV are incredibly similar activities if you have a nice lawnmower.
I actually think there are mental benefits to cutting the lawn. There is something zen about slowly reducing the grass in an ordered manner.
I've had very similar thoughts about lawn-cutting, but hadn't related it to technology-related trends!
I couldn't agree more. In a broad sense, it's like we've lost contact with our own lives, communities, and cultures. I don't think those things are "dead", but they mean things now which are totally different than what they meant even only a couple of decades ago, and the people living through it sort of know that and kind of say it out loud sometimes even, but can't really wrap their heads around it at the same time and continue to ape the old behaviours (no offence, anyone, I do it too).
"Traditions" live on in a sort of zombie state... we cut the grass, we present the gift cards. It's an odd and fun(ny) moment to be alive.
I'm not disputing that people waste inordinate amounts of time running out the clock before they shuffle off this mortal coil. (cough every X demograph reacts to Y video cough).
Agreement to eat said head apparel is predicated upon "infectious energy" (i.e. quality) - NOT success. I'll draft up a more officious document later.
Note: I am the sole arbiter of what constitutes quality.
You also seem to be the sole arbiter of what constitutes worthy activities.
> It's been a long time since "quality" and "success" have been decoupled.
I think so, too.
I guess quality was a property of interest in the old days, because the path e.g. for commercial music was: Maximize profit -> Maximize sales -> Maximize what the target audience likes -> Maximize quality.
For TikTok etc. they bypass the market sales stuff and replace it by an 'algorithm', that optimizes for retention, which is tightly coupled to ad revenue. I imagine the algorithm as a function of many arguments.
Just relying on quality is an inefficient approximation in contrast to that.
The Art of Zen and Motorcycle Repair will teach anyone all they ever need to know about quality.
We used to have some "radio station" in shoutcast during the 90s. It was very local and geeky, but had a couple of listeners. It was kind of a podcast
Good point. I absolutely abhorred reaction videos until I found a channel that works for me. A guy who reacts to videos about ultra luxury mansions. I find it interesting hearing an expert in the field point out things that most of us wouldn't think of. Plus he's got a great sense of humor.
What blew my mind was that there would be a market to watch other people play video games. You never know what catches on.
> Go back to the 1990s (even probably the 2000s?) and ask literally anyone what they think of the idea of people spending huge quantities of their time on this mortal plane watching videos of other people talking about how they do their make up and pick their outfits.
So like supermarket magazines in video form? I think people would get it.
To me, unboxing videos are documentation, not entertainment. When I need to know exactly what's in a box and how it's packaged, an unboxing video is the only source of that information.
What's in the box is almost always on the box, and how it's packaged seems to be very useless information.
As someone who's struggled to really get into podcasts, I'm convinced that most people who enjoy podcasts don't really actively listen to them, they just like that extra bit of noise in the background while they do something else, with the added bonus that they might just listen at just the right time to pick up some interesting factoid.
Don't get me wrong, I like podcasts, Fall of Civilizations being a personal favorite of mine, it's just that my desire to try and actively listen to them requires that I carve out an hour (or much longer in the case of FoC, worth it) of my free-time to eliminate any distractions and focus. Pretty hard to do when there are so many other things I could be doing with my time besides sitting there trying my best to listen.
Anyway, I bring it up because I'm convinced that the people promoting AI-podcasts are mostly made up of the aforementioned people who just listen to them for noise.
I used to listen to podcasts during my commute, when they took the place of listening to NPR or talk radio. Now that I work from home full time, I listen to them when I'm doing dishes or similar chores, going on walks around the city, and when I'm doing any significant amount of driving. I listen to a combination of comedy/history (love The Dollop), politics, and sci-fi commentary.
For the most part, I'm listening pretty actively, but if I'm just sitting there listening without a fairly mindless activity going on, I'll get distracted pretty quickly and find myself looking at my phone...
I don't get them either since the hosts are typically "regular people" talking about often complicated subjects that they are by no means domain experts in.
I find the format of "Dumdum Host A read some articles about something last night and Dumdum Host B asks questions about it" especially grating, unless it's purely opinion-driven and then I still probably won't care unless I read about the hosts and find out they are probably people with opinions worth listening to.
I'd rather read a book or be with my own thoughts without having them be even more crowded by some randos telling me stuff they think they know
Not all podcasts are Joe Rogan
Oh yeah, i definitely hate it when uninformed people try to make podcasts on specialized topics they obviously know nothing about. I went searching for niche physics podcasts (anything besides space stuff please!) And the amount of literal high schoolers and undergraduates attempting to explain things they're years away from even seeing in class is painful, most admit that they know nothing and are literally free-associating over a Wikipedia article. I give them a pass though since they have no listeners and are usually young but...
I exclusively listen to domain expert podcasts. History experts, policy/economics experts, aforementioned physics experts. Literally event i listen to is hosted by at least one person who is a domain expert. They can branch out and interview experts in other domains but there's always an expert. I honestly can't say (besides the high schoolers which i searched out) the last time i've even test listened to a podcast hosted by semi-charismatic article readers.
I'd recommend you search reddit for lists of podcasts in your desired area. App recommendations are all shit and there's many posts out there
> I don't get them either since the hosts are typically "regular people" talking about often complicated subjects that they are by no means domain experts in.
Huh? Most podcasts I listen to feature either experts in the field in question or very skilled journalists compiling the opinions of experts.
> Don't get me wrong, I like podcasts, Fall of Civilizations being a personal favorite of mine, it's just that my desire to try and actively listen to them requires that I carve out an hour (or much longer in the case of FoC, worth it) of my free-time to eliminate any distractions and focus.
I don't see why that should be true. Podcasts are great when coupled with mindless work like mowing the lawn, weeding the garden, stacking firewood, walking to the shop, driving etc. You can get virtually 100% out of a podcast while performing tasks like this.
I figured people listened to them while commuting. It seemed the best fit: lots of time, mostly little attention needed (except when you alight or are driving in adverse conditions).
I don't understand enjoying podcasts while just using them for background noise. Maybe because I actually listen, and most podcasts I listen to seemed to have an engaged audience. But then I like listening to people talk about philosophy, politics, sports, true crime, science, history.
I agree. I think I listen to podcasts just for a bit of "social noise" when I'm doing something by myself and for occasionally picking up on something new I hadn't heard of before. For pure information content, I think they're actually very poor. It's not unlike listening to an old AM radio talk show. Hosts repeat themselves, engage in banter, and often oversimplify topics for the sake of a narrative.
I also think AI podcasts could become popular one day for people who just want some background noise and bits of trivia every once in a while. I would argue that a lot of YouTube channels I sometimes have in the background just summarize Wikipedia articles and don't have much of a personal touch anyway, so an AI could do the same thing.
I switched to improvised pods for this reason. Podcasts are for doing the dishes and mowing the lawn and playing factorio. I don't turn on anything with 'meat' unless I have a long car trip ahead of me. I just am not able to sit still and only listen without feeling like I could be doing something else, and when I'm doing something else I'm gonna start tuning out eventually.
Hey Riddle Riddle, Hello from the magic tavern, Artists on artists on artists on artists are my picks for now.
There is nothing you are supposed to be doing with free time, hence free time.
Tbh tho, there is nothing that we are supposed to be doing in this world rn anyways, nobody was born to work any of these jobs, we didn't evolve to flip burgers.
This is the old world, don't place too much stake in it, everything will change and then what?
If you want to listen to a podcast, do it. If you want to do something else, do that.
Don't do what you don't want to do, that's all there is too worry about.
This is partially true for me, but when I listen to podcasts it’s generally in what would otherwise be dead time for me such as when grocery shopping or exercising.
Having this strong of an opinion on something you admittedly don't understand (the appeal of podcasts) is not a great choice. Maybe ask people instead of making up stuff.
I have very strong opinions on lots of things I can't understand why people like doing, I don't need to understand someone's satisfaction before I determine how I feel about what they are driving satisfaction from. That's ridiculous.
I think people can't be alone with their own minds so they fill every second of there quiet time with someone else's thoughts rather than their own.
> I'm convinced that most people who enjoy podcasts don't really actively listen to them,
Or you know not everyone is the same. Just because you struggle with something doesn’t mean that it is not easy and effortless for others.
I can’t dribble a basketball and walk at the same time, still I won’t make the claim that everyone who claims to enjoy playing basketball is somehow fudging it.
[dead]
I have probably written this comment about a dozen times on HN already, but: I agree completely, because people don't listen to podcasts purely for information, they listen for information plus community, personality, or just a basic human connection.
If you're a content creator today, the best thing you can do to "AI-proof" your work is to inject your personality into it as much as possible. Preferably your physical personality, on video. The future of human content is being as human as possible. AI isn't going to replace that in our lifetimes, if ever.
> The future of human content is being as human as possible. AI isn't going to replace that in our lifetimes, if ever.
I have been working on machine learning algorithms for a long time. Since the time when telling someone I worked on AI was a conversation killer, even with technical people.
AI's are going to understand people better than people understand people. That could be in five years, maybe - many things are happening faster than expected. Or in 15 years. But that might be the outside range.
There is something about human psychology where the faster something changes, the less we are aware of the rate of change. We don't review the steps and their increasing rate that happened before we cared about that tech. We just accept the new thing that finally gets our attention like it was a one off event, instead of an accelerating compounding flood, and imagine it isn't really going to change much soon.
--
I know this isn't a popular view.
But what is happening is on the order of the transition to the first multi-cellular creatures, or the first bodies with specialized cells, the first nervous systems, the first brains, the first creatures to use language. Far bigger than advances such as writing or even the Internet. This is a transition to a completely new mode for the substrate of life and intelligence. The lack of dependency on any particular substrate.
"We", the new generation of things we are building, will have none of the limits and inefficiencies of our ancient slow DNA-style life. Or our stark physical bottlenecks on action, communication, or scaling, and our inability to understand or directly edit our own internal traits.
We will have to face many challenges in the coming years. It can't hurt to mindfully face them earlier than later.
What machine learning algorithm have you worked on that leads you to believe they are capable of having a rich internal cognitive representation anywhere close to any sentient conscious animal?
If you pick any well performing AI architecture, what would lead you to believe that they are not capable of having a rich internal cognitive representation?
The Transformer, well... transforms, at each layer to produce a different representation of the context. What is this but an internal representation? One cannot assess whether that is rich or cognitive without some agreement of what those terms might mean.
LLMs can seemingly convert a variety of languages into an internal representation that encompasses the gist of any of them. This would at least provide a decent argument that the internal representation is 'rich'
As for cognitive? What assessment would you have in mind that would clearly disqualify something as a non-cognitive entity?
I think most people working in this field who are confident feel that they can extend what they know now to make something that looks like a duck, walks like a duck, and quacks like a duck. If that is achieved, on what basis does anyone have to say "But it's not really a duck"?
I'm ok with people saying AI will be never able to perform that well because it doesn't have X, as long as they accept that if it does, one day, perform that well they accept that either X is present, or that X is not relevant.
If you think we're only our observable behaviors or that is the only relevant thing to you then I don't think it's worth getting into this argument. Consider this excerpt from https://scottaaronson.blog/?p=7094#comment-1947377
> Most animals are goal-directed, intentional, sensory-motor agents who grow interior representations of their environments during their lifetime which enables them to successfully navigate their environments. They are responsive to reasons their environments affords for action, because they can reason from their desires and beliefs towards actions.
In addition, animals like people, have complex representational abilities where we can reify the sensory-motor “concepts” which we develop as “abstract concepts” and give them symbolic representations which can then be communicated. We communicate because we have the capacity to form such representations, translate them symbolically, and use those symbols “on the right occasions” when we have the relevant mental states.
(Discrete mathematicians seem to have imparted a magical property to these symbols that *in them* is everything… no, when I use words its to represent my interior states… the words are symptoms, their patterns are coincidental and useful, but not where anything important lies).
In other words, we say “I like ice-cream” because: we are able to like things (desire, preference), we have tasted ice-cream, we have reflected on our preferences (via a capacity for self-modelling and self-directed emotional awareness), and so on. And when we say, “I like ice-cream” it’s *because* all of those things come together in radically complex ways to actually put us in a position to speak truthfully about ourselves. We really do like ice-cream.
> And when we say, “I like ice-cream” it’s because all of those things come together in radically complex ways to actually put us in a position to speak truthfully about ourselves. We really do like ice-cream.
Ok, now prove this is true. Can you do so without invoking unobservable properties? If you can, then observable is all that matters, if you cannot then you have no proof.
Do I seriously have to prove to you that you like ice cream? Have you tried it? If you sincerely believe you are a husk whose language generation is equivalent to some linear algebra then why even engage in a conversation with me? Why should I waste my time proving to you a human that you have a human experience if you don’t believe it yourself?
You don't need to prove to me that I like ice cream. You need to prove to me that you like ice cream. That you even have the capacity to like. Asserting that you have those experiences proves nothing since even a simple basic program 10 print "I like Ice Cream" can do that.
How can you reliably deny the presence of an experience of another if you cannot prove that experience in yourself?
I actually don’t need to prove to you that I’m more than a BASIC program. I mean listen to yourself. You simply don’t live in the real world. If your mom died and we replaced her with a program that printed a bunch of statements that were designed to as closely mimic your conversations with her as much as possible you wouldn’t argue hey this program is just like my mom. But hey maybe you wouldn’t be able to tell the difference behind the curtain so actually it might as well asbe the same thing in your view, right? I mean who are we to deny that mombot is just like your mom via an emergent pattern somewhere deep inside the matrices in an unprovable way /s. Just because I can’t solve the philosophical zombie problem for you at your whim to your rigor doesn’t mean a chatbot has some equivalent internal experience.
I'm not claiming that any particular chatbot has an equivalent experience, I'm claiming there is no basis beyond its behaviour that it does not.
With the duplicate mother problem, if you cannot tell then there is no reason to believe that it is not a being of equivalent nature. That is not the same as identity, for a layman approach to that viewpoint, see Star Trek: TNG, Season 6, Episode 24. A duplicate Will Riker is created but is still a distinct entity (and one might argue, more original since has been transported one fewer times). Acting the same as is not the same as being the same entity. Nevertheless it has no bearing on whether the duplicate is a valid entity in its own right.
You feel like I'm not living in the real world, but I am the one asking what basis we have for knowing things. You are relying on the presumption that the world reflects what you believe it to be. Epistemology is all about idetifying exactly how much we know about the world.
Ok you can have your radically skeptic hard materialist rhetoric. I just don’t take it seriously, and I don’t think you do either. It’s like those people who insist there is no free will and yet go about their day clearly making choices and exercising their free will. If you want to say technically everyone might as well be a philosophical zombie just reacting to things and your internal subjective experience is an illusory phenomenon, fine you can say that as much as you want. In turn I’ll just give up here because you don’t even have a mind that could be changed. I can sit here and claim you’re the equivalent of a void that repeats radically skeptic lines at me. Maybe a sophisticated chatbot or program. Or maybe you’re actually a hallucination since I can’t prove anything really exists outside of my sense. In which case I’m really wasting my time here.
Well I'm a combatibilist, So I certainly believe in free will. I also will accept any entity that consistently acts as if it has a will that it does actually have that. That's has always been my point, treating things as what they appear to be is the only rational approach when you cannot prove or disprove the existence of the property in question.
It follows from that that you cannot exclude something that appears to have a property if you cannot prove it doesn't or even prove it if it does.
Fair enough. I wouldn’t say a program is acting with a will of its own just because it’s trained to respond to questions in a human like way. That doesn't even say anything about its capacity to have a will. Language is a tool that can convey internal state, not the thing itself.
This is basically the Turing test, and like the Turing test it undervalues other elements which can allow for differentiation between “real” and “fake” things. For example - if we can determine that a thing that looks, walks, and quacks like a duck, but doesn’t have the biological heritage markers (that we can easily determine) then it won’t be treated as equivalent to a duck. The social desire to differentiate between real and fake exists and is easily implementable.
In other words: if AIs/robots ever become so advanced that they look, walk, and talk like people, I expect there to be a system which easily determines if the subject has a biological origin or not.
This is way down the line, but in a closer future this will probably just look like verifying someone’s real world identity as a part of the social media account creation process. The alternative is that billion dollar corporations like Meta or YouTube just let their platforms become overrun with AI slop. I don’t expect them to sit on their hands and do nothing.
> AI's are going to understand people better than people understand people.
Maybe, but very little of the “data” that humans use to build their understanding of humans is recorded and available. If it were it’s not obvious it would be economical to train on. If it were economical it’s not obvious that current techniques would actually work that well and by definition no future techniques are known to work yet. I’m not inclined to say it will never happen but there are a few reasons to predict it’ll prove to be significantly harder to build AI that gets out of the uncanny valley that it’s currently in.
You are describing the current state of AI as if it were a stable point.
AI today is far ahead of two years ago. Every year for many years before that, deep learning models broke benchmark after benchmark before that breakout.
There is no indication of any slow down. The complete reverse - we are seeing dramatic acceleration of an already fast moving field.
Both research and resources are pouring into major improvements in multi-modal learning and learning via other means than human data. Such as reinforcement learning, competitive learning, interacting with systems they need to understand via simulated environments and directly.
> You are describing the current state of AI as if it were a stable point.
No I’m not, I’m just not assuming that the S curve doesn’t exist. There’s no guarantee that research results in X orders of magnitude of improvement that will result in AI being better at understanding humanity than humans in the 5 to 15 year timeframe. There’s no guarantee that compute will continue to consistently grow in volume and lower in price, and a few geopolitical reasons why it might become rarer and prohibitively expensive for some time. There’s no reason to assume capital will stay as available to further both AI techniques and compute resources should there be any sign that investments might not eventually pay off. There’s also no reason to assume the global regulatory environment will remain amenable to rapid AI development. Maybe the industry threads all these needles, but there’s good reason to predict it’ll prove doesn’t.
They are not using AI correctly to create such models. I'm not sure I want AGI right away or even at all, so I'm keeping my epiphany close for now but in the current field of AI nothing er wany will come of this bc it's not the right way.
As soon as the incredibly obvious, far too obvious realization is had, AI will make huge, tremendous leaps overnight essentially. Til then, these are just machine like software, the best we've ever made but nothing more than that.
This makes me think about attention span. Scenes in movies, sound bites, everything has been getting shorter over the decades. I know mine has gotten shorter, because I now watch highly edited and produced videos at 2x speed. Sometimes when I watch at 1x speed I find myself thinking "why does this person speak so slowly?"
Algorithmic content is likely to be even more densely packed with stimuli. At some point, will we find ourselves unable to attend to content produced by a human being because algorithmic content has wrecked our attention span?
Sorry but I don’t think this is much evidence of anything. The point at which an AI can imitate a live-streaming human being is decades away. By then, we will almost certainly have developed a “real Human ID” system that verifies one’s humanity. I wrote about this more here:
https://news.ycombinator.com/item?id=42154928
The idea that AI is just going to eat all human creative activity because technology accelerates quickly is not a real argument, nor does it stand up to any serious projections of the future.
AI is already eating their way up the creative ladder, this is 100% irrefutable. Interns, junior artists, junior developers, etc are all losing jobs to AI now.
The main problem for AI is it doesn't have a coherent creative direction or real output consistency. The second problem is that creativity thrives on novelty but AI likes to output common things. The first is solvable, probably within 5 years, and is going to hollow out creative departments everywhere. The second is effectively unsolvable, though you might find algorithms that mask it temporarily (I'm not sure if this is any different than what humans do).
We're going to end up with "rock star" teams of creative leads who have more agility in discovering novelty and curating aesthetics than AI models. They'll work with a small department comprised of a mix of AI wranglers and artisans that can put manual finishing touches on AI generated output. Overall creative department sizes are probably going to shrink to 20% of current levels but output will increase by 200%+.
How can you both think AI will do a soulless garbage job and that it will displace all the creative people who put their blood sweat and tears into doing art.
If you think it can be 20 times easier to make a movie, then it seems to me that it would be 20 times less impactful to make a creative work, since the market should quickly react to creative success by making a ton of cheap knockoffs of your work until the dead horse is so thoroughly beaten that it's no longer even worth paying an AI to spit out cheap knock-offs
> How can you both think AI will do a soulless garbage job and that it will displace all the creative people who put their blood sweat and tears into doing art.
They don't, that's why they're saying 20% of creative departments will remain. The part that will go is the part that's already soulless - making ads in the style of the current trendy drama series or what have you.
> If you think it can be 20 times easier to make a movie, then it seems to me that it would be 20 times less impactful to make a creative work, since the market should quickly react to creative success by making a ton of cheap knockoffs of your work until the dead horse is so thoroughly beaten that it's no longer even worth paying an AI to spit out cheap knock-offs
That seems pretty backwards, given Jevons paradox. The ease of writing knock-off fanfiction didn't mean people stopped writing novels. The average novel probably has a lot less impact now than in the past, but the big hits are bigger than ever.
How do you know it is decades away? A few years ago did you think LLMs would be where they are today?
Is it possible you’re wrong?
Of course it’s possible I’m wrong. But if we make any sort of projection based on current developments, it would certainly seem that live-streaming AI indistinguishable from a human being is vastly beyond the capabilities of anything out today, and given current expenses and development times, seems to be at least a few decades in the future. To me that is an optimistic assumption, especially assuming that live presence or video quality will continue to improve as well (making it harder to fake.)
If you have a projection that says otherwise, I’d be glad to hear it. But if you don’t, then this idea is merely science fiction.
Making predictions about the future that are based on current accelerating developments are how you get people in the 1930s predicting flying cars by 2000.
You imply that the technological development will stop, but that's not what happened to flying cars - they do and could exist. Since the 1930s the technology didn't stop developing - aircraft went jet powered, supersonic, to the edge of space, huge, light, heavy, more efficient, more affordable, safer; cars went faster, more reliable, safer, more efficient, bigger, smaller, more capable, self driving; fuel got more pure, engines got more power per kilogram, computer aided design and modelling of airflow and stress patterns was developed, stronger lighter materials were developed; flying cars have been built:
Klein Vision AirCar, 2022: https://www.youtube.com/watch?v=5hbFl3nhAD0
Predecesor, the AeroMobil from 2014, from prototypes starting 1990: https://en.wikipedia.org/wiki/AeroMobil_s.r.o._AeroMobil
The Terrafugia Transition 'roadable aircraft' from 2009: https://en.wikipedia.org/wiki/Terrafugia_Transition
Moller SkyCar, of course, which never got to free flight, but was built and could fly.
The problems are regulatory, cost, safety, massive amounts of human skill needed, the infrastructure needed, the lack of demand, etc. A Cessna 182 weighs <900 Kg, a Toyota Corolla weighs >1400 Kg and has no wings, no tail. But if we collectively wanted VTOL flying cars enough to put a SpaceX level of talent and money at them, and were willing to pay more, work harder, for less comfort and more maintenance, we could have them. Bit like Hyperloop; a low-pressure tunnel with carriages rushing through it is not impossible, but it's got more failure modes, more cost, more design difficulties and almost no benefits over high speed train and maglev.
I didn't mean to imply that, but that's my fault for just quickly using it as an example.
I do think there will be some slowdown of the technological development, but I think the situation will be similar to flying cars: regulations, social behavior, etc. will prevent AI from simply devouring everything. Specifically, I expect there to be a system which verifies humanity, and thus popular content will ultimately end up being verified-as-real content.
More on that in this other comment: https://news.ycombinator.com/item?id=42154928
I don't think regulations and social behaviour could stop a Skynet style "AI go FOOM" future. It might be able to stop LLM filler from covering the internet, but given how it can't stop carbon dioxide filler from the atmosphere and "AI text" is less measurable, less easy to stop, and has less clear consequences, I'm not sure I'd bet on it.
Possibly in the sense of going to a small server where ID is verified and hiding away on Discord from the rest of the internet. That would be more like building a bunker to hide from the flying cars, rather than flying cars never happening.
Do 1970 instead, over50 years of stagnation.
For many people current AI created videos are already confused for real videos and vice versa.
When you follow up on technology by browsing HN, and see the latest advancements, its easier to see or hear the differences, because you know at what to look.
If I see on tv some badly encoded video, especially in fog or water surfaces, it immediately stands out, because I was working with video decoding during the time it was of much less quality. Most people will not notice.
People are judging AI by what abusers of AI put out there for lols and what they themselves can wring out of it. They haven't yet seen what a bunch of AAA professionals with nothing to lose can build and align.
Billions of dollars and all of Silicon Valley's focus has been spent over the last 2 years trying to get AI to work, the "AAA professionals" are already working on AI and I still have yet to see an AI generated product that's interesting or compelling.
The cornerstone of genAI hype is "AI for thee, not for me"
it's filled with tech people who are fucking morons thinking that everyone else is really dumb and loves slop and their algorithm will generate infinite content for the masses, making them rich. yet they don't consume it themselves, and aren't smart enough to recognize the cognitive dissonance
anyone who thinks unsupervised AI content is going to replace [insert creative output here] shouldn't be in the HN comment section, they should be using ChatGPT to generate comments they can engage with
instead they're going to get mad about me calling them a fucking moron in this comment. which, like, why get mad, you can go get an LLM to generate a comment that's much nicer and agrees with you
> anyone who thinks unsupervised AI content is going to replace [insert creative output here] shouldn't be in the HN comment section, they should be using ChatGPT to generate comments they can engage with
Actually, their handlers should just be better monitoring their internet usage
Do you think the people who make TikTok or Clash of Clans (or, hell, the Daily Mail) use their own products?
uh yes? Was this supposed to be some clever "gotcha!" that ended up being really stupid? Might want to run your clever comments by ChatGPT before you post next time chief
> uh yes?
Well, you're wrong, I know some of them.
oh word? i know a guy at Microsoft who doesn't use Excel, that means nobody at Microsoft ever uses Excel and they're just peddling crap to the dumb masses
Your argument is weak.
the fact that you're engaging w/ a random human on the internet proves that my argument is strong
if you don't like it, go talk to Claude about it
Claude will even use proper punctuation & grammar which i'm not doing. it's LITERALLY an objective improvement over this comment. why the hell are you engaging with this crap?
No AAA professionals in the entertainment profession have built a gen AI model up through their own vision that I'm aware of. AAA Nerds build the next version to distract away from flaws of the last and hope the api will get heavy use. I wouldn't expect output to be compelling with that attitude. I expect repulsive or parody clips to continue until some great creatives feel like building their replacements properly.
Yes, this seems possible. I only wonder if it is too fragile to be self perpetuating. But we're here after all, and this place used to be just a wet rock.
I find your ideas intreguing, and I wish to subscribe to your newsletter
I just took a poll about art that had 50 examples and you had to choose which were human and which were not - I did very well, bc I'm awesome but apparently it shook the art world to its core, many of the professionals have no idea how to tell the difference/can't tell the difference.
The elements that identified a painting as human to me had nothing to do with appearance but everything to do with feeling.
Be passionate about what you present - if you can't be, don't bother. Passion is what will distinguish us from the AI - stuff like personality is almost immediately replicated, vocal inflections, funny comments, unusual spoken delivery - that's all within the realm of AI capability. Talking about micro blogging like it's God's gift to the world, not so much an AI thing, it won't feel right with AI.
Agreed. I finally got around to listening to a NotebookLM generated podcast this weekend, and found it absolutely unlistenable.
For some reason the LLM seemed to latch on to the idea that one host should say something, and the other host should just repeat a key phrase in the sentence for emphasis -- and say nothing else, and they'd do it like multiple times in a sentence, over and over again throughout the entire thing.
Slightly less weird -- but it seems like the LLM caught on that a good narrative structure for a two-host podcast is that one host is the 'expert' on the topic, and the other host plays dumb and ask questions. Not an unreasonable narrative structure. Except that the hosts would seamlessly, and very weirdly, switch roles constantly throughout the podcast.
And ultimately the result was just a high level summary of the article I had provided. They told me in the intro and the outro about the interesting parts they were going to dive into, but they never actually got around to diving into those parts.
I tried it with something fairly abstract about a decision I was working on making. Fed it a bunch of information, details about me, my background, the factors I'm considering in the decision and the impact of getting it right/wrong.
It was interesting. I certainly wouldn't say it was useless, I think the contrived dialogue actually touched on some angles I hadn't considered and I think it was useful. Not in a 'oh shit it's clear to me now' kind of way but it definitely advanced my thinking.
I find it as listenable as the podcasts the people I'm around play in their cars during roadtrips
To me that's absolutely unlistenable, but to them its interesting and engaging. I find NotebookLM replicates that perfectly. Its not at all the issue that the OP encountered with the Hawaii news service, as those lacked tone and pronounciation, which NotebookLM would not.
Regarding the outcome being a long winded summary, yeah thats what I see about the aforementioned 2 hour podcasts as well. They take the intro paragraph of a wikipedia article, pretend that the topic is a novel mystery they just discovered through hours of scouring microfilm at a municipal library, and then interject each other every other word with nonsense, before getting back to the point for just one sentence.
> For some reason the LLM seemed to latch on to the idea that one host should say something, and the other host should just repeat a key phrase in the sentence for emphasis -- and say nothing else, and they'd do it like multiple times in a sentence, over and over again throughout the entire thing.
cough https://en.wikipedia.org/wiki/ELIZA
Hmm, does Google not use the Claude or ChatGPT API for it? I still don't hear of people chatting with Gemini nearly as often as with the other two.
Edit: looks like the subreddit is still called Bard. Well played, Internets. https://www.reddit.com/r/Bard/comments/1g0egad/gemini_vs_not...
There are a lot of "uh huh", "yeah", "agree" types of statements in odd places as well.
> If you can find an LLM+TTS generated podcast with even a FRACTION of the infectious energy
this is the wrong standard. there's a long tail of many many many many topics that a Radio Lab will never have the resources or critical mass or staff or turnaround time to cover. AI can serve the long tail like nothing you have ever seen. (because humans are expensive, and deservedly so, but our needs and wants are infinite)
You're not wrong, but your argument seems to boil down to "something is better than nothing" and I suppose I disagree. If there's not a podcast about a particular topic that I'm interested in, I would rather go find a curated expose of that topic via a different medium, such as longform journalism.
I think you make a good connection with the long tail. That will definitely be one of AI's triumphs. People who don't understand this, don't understand what's been happening to media production and consumption over the last 40 years.
Infectious energy is something I expect faked relatively easily, though I don't know your examples and doing it in AI might be a "first 90%" situation just like self driving cars; for me the problem is that they're fairly mediocre at the actual script — based on me putting a blog post I wrote into one and listening to what came out.
Given how many podcasts exist, I think you need to be at least 2 standard deviations above mean to even get noticed, 3 to be a moderate success, and 4 to be in the charts.
I'd guess AI is "good enough" to be 1 above average, as the NotebookLM voices sound like people speaking clearly and with some joy into decent microphones in sound isolating studios.
I probably should've clarified that by infectious energy I wasn't so much referring to the vocal aspect as I was the overall quality, interaction between the hosts, and pithiness / wit.
Having experimented with many LLMs (mixtral, sonnet, ChatGPT, Llama, etc.), the coherence is for the most part on point, but their capacity for novelty has been found wanting irrespective of how I tuned the top_k, temperature, or prompts.
That being said, I've seen some very impressive examples of style transference even conveying emotional range in some of the SOTA TTS systems.
Excuse me for being an old geezer but at least AI bots don't tend to pepper their sentences with frequent utterances of 'like'. I don't normally find this speech mannerism annoying, and I do it myself too, but when 'like' is overused I switch off.
For this reason I don't listen to the otherwise highly entertaining CineFix podcast. Example: the recent episode discussing Kill Bill vol 1 contains 624 utterances in 75 mins (8.3 per min). IANAL I know.
Ha that’s not what IANAL means but perhaps should be
I've yet to hear an LLM-generated podcast about an article that wouldn't have been better if the AI had simply read the article aloud.
Someone in 2020: "If you can find a computer program that can generate an image with even a FRACTION of the aesthetic appeal of human created art...."
Give it time. Progress is fast.
> Give it time. Progress is fast.
Why pick 2020 as your starting point? That is simply around the time the current set of techniques came about.
We had generative art back in the late 90's - my screensaver has been generative art for over 20 years now.
Obviously generative art has come a long way but people have been working on various approaches to it for at least 30 years.
From 2020 until now, we've gone from crude blurry or clearly generative artefacts to being able to create full professional illustrations based upon textual prompts. That is huge. Classic generative art techniques look like cave paintings compared to what the latest image generation models put out (and I'm not talking about "AI slop" type stuff that DALL-E does).
Similarly, tools could fabricate podcasts years ago that sounded terrible. Now we have NotebookLM doing a "reasonable" job with two cliched-sounding "hosts". In a few years, will they potentially be able to create something akin to a professionally produced podcast given some smart prompting? The progress made so far points to yes, and I haven't seen any evidence so far to be pessimistic about it happening.
Can current techniques be scaled/improved/optimized to do this or do we need new techniques?
It took 30 years in the generative art world to move from cave paintings to the level that we have today because we needed new techniques.
For podcasts we are at the cave paintings level.
If we can get to professional level quality podcasts with the current techniques then we might only be a few years away.
I think it is more likely we will need new techniques which puts us potentially decades away.
If we look at LLMs the improvements over the last 18+ months since gpt4 was released have been minor despite incredible levels of investment.
> "being able to create full professional illustrations based upon textual prompts. That is huge. .. (and I'm not talking about "AI slop" type stuff that DALL-E does)"
Then what are you talking about, where can I get it?
“Generate” does not have the same connotation that “create” does. Are they really that interchangeable?
I've heard so many people say that podcasts are just something they have playing in the background while they do other things that I have no trouble believing that they'd play an AI podcast in the same way.
I personally would not bother with AI generated podcasts, they're such low bar, why waste time where there's so much other great content to catch up on? But I think you may be right, I wouldn't be surprised if people take them in with no fuss. But then what do I care? What I care most is that they'll pollute the search space. I'd filter out all GenAI content if I had the option and Im guessing that will become an option soon.
The value in AI podcasts at the moment isn't in replacing human content, but in filling the niches that human content just doesn't cover. Doesn't matter if it's not the best podcast ever, when it's literally the only podcast on this planet discussing the topic you want to listen too.
People think they want to use the Minority Report computer interface because it looks cool and advanced but they don't put the least amount of thought into it and realize it's terribly impractical. Our arms would get tired very quickly. A mouse resting on a table isn't further from ideal just because it was invented earlier.
Fooling people with the promises of AI is pretty simple. People are easy to fool. They like shiny objects.
When TV was first introduced, the first broadcasts were people standing by a mic reading a radio script.
AI podcasts aren't going to be drop in replacements for exactly the length, frequency, personalities, topics, and style of current podcasts. Claiming that's a fault with AI podcasts just indicates a lack of imagination.
The main reason im not concerned about AI-based entertainment is the same reason I watch human chess players. It's not only about technical capabilities. I can't explain fully why though..
I've heard amazing catchy songs from Suno and Udio. So much so they're still stuck in my head as earworms several months later. If they'd been streaming on youtube or spotify I wouldn't have given it a thought that they might be AI generated.
So, I can certainly imagine a podcast doing the same to some degree. Maybe not a podcast where AI wrote the script, but, a podcast where AI read a story dramatically doesn't seem too far off or, easier, a podcast that read news to me.
We had this recently, yes.
https://news.ycombinator.com/item?id=41693087
It's kind of horrible.
Again - think of where we were two years ago. I never understand this hubris people have to think AI can never do X while being proved repeatedly wrong.
Everybody trots out this argument.
GPT styled LLMs were introduced back in 2018 so SIX years ago.
Have they gotten more COHERENT? Absolutely. Is coherence the same thing as NOVELTY? NOT EVEN REMOTELY. I've played with markov chains in the 90s that were capable of producing surprising content.
Unless there is a radical advancement in the underlying tech, I don't see any indication that they'll be capable of genuine novelty any time in the near future.
Take satire for example. I have yet to see anything come out of an LLM that felt particularly funny. I suppose if the height of humor for you is Dad jokes, reddit level word punnery, and the backs of snapple lids though that might be different.
If you have a particular style of witty observational humor that you prefer, providing the model some examples of that will help it generate better output. It's capable of generating pretty much anything if you prompt it the right way. For truly nuanced or novel things, you have to give it a nucleus of novelty and axis of nuance for it to start generating things in your desired space.
If I tell it exactly what I want to hear, where's the surprise? Is there any novelty there?
> I don't see any indication that they'll be capable of genuine novelty any time in the near future.
That is like saying the plane invented by Wright brothers will never go the moon.
Two years ago we had ChatGPT and Midjourney. Now... we also have those?
The accelerationism argument made a little bit of sense two years ago, but now, after two years of marginal improvements at best? Really?
Because the problem is that "AI generated" content is vapid slop. The problem is that most "human generated" content is also vapid slop.
My problems on the modern internet are simply that I can't find interesting things through the slop, period--human or AI generated.
It will be interesting to see if "curation" can actually find a business model this time.
I totally agree - LLMs may be great for spitting out content, but they just can’t quite replicate the energy, personality, and creativity of an actual podcast. The human touch is the irreplaceable component.
It doesn't quite stop at podcast hosts either. The simple fact that LLMs will never truly be capable of replacing software developers was one of the core principles behind our team developing The Ac28R - A completely new type of AI capable of bulletproof industrial software. It won’t host a podcast anytime soon, but if you need flawless code for complex projects, it's got you covered!
I think we will get a surprising amount of AI generated content in the future. During the first year of the Urkain invasion there was an enormous amount of AI voiced and scripted video content on YouTube. I think AI content will take over in the easy parts first. And over time take up more and more views.
Do you not use autocorrect?
I only discovered https://notebooklm.google.com/ today, as an experiment i threw in a dry EU directive, https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CEL... and got it to generate a podcast that was 20 mins long and was a good intro to the directive and natural to listen to. It won't replace podcasts that require creativity but as a tool to summarize info into a digestible format I'll most definitely be using it again.
Businesses/CEOs want to show profitability by spending less on human employees. Human consumers don't want to lose the human touch. Will be really interesting to see how many of the consumer facing AI startups actually make it.
Many don't even care if they make it in the long run. Making it is cashing in in the short term, screwing the investors.
I've never found anyone telling me their opinions to be infectious - I can't even imagine using that word to describe a podcast tbh. Its a form of media I've never understood - they script the podcasts, they all kno the conversation ahead of time and just go thru the motions via voice with strangers that can either agree or disagree - only the extremes will ever interact with the podcasters, it's just so fake I can't even do it.
Infectious energy. Interesting.
How would you go about eating said stovetop hat?
Would you eat it raw or would you have to boil it down to soften it up and then drink/eat its hat soup?
Thank you for focusing on the real point of my post. :)
I was thinking I'd just unhinge my jaw in the style of 90s Reach Toothbrush commercials or like a human pez dispenser.
Cook it like stovetop stuffing, I'd imagine....
Isn't it a stovePIPE hat?
I do enjoy some of the podcasts mentioned in this thread, but struggle to find good non-American ("foreign"?!) podcasts to listen to. Similarly to finding good quality non-US film and television, it can be hard to locate but I greatly enjoy it.
If anyone has suggestions for podcast collation sites that are for non American content that would be fantastic.
I think it's quite easy for AI to surpass the median podcast. The bar is fairly low there.
I guess that matches the state of AI in general, better than a novice; worse than an expert.
We will have to wait and see what future AIs deliver. Insight and nuance is what I look for in media like that, that's a much harder nut to crack.
same is true for AI music, AI art, AI articles to some extent. most communication is about human connection. you remove it and the communication becomes worthless. AI is great for communication that has less personal involvement like drafting short professional emails. that's also why all the AI agent craze seems problematic: https://medium.com/thoughts-on-machine-learning/langchains-s...
> Slightly tangential, but this is the reason I'm baffled why people think that AI-driven podcasts would ever be worth listening to
the only real people that believe in that that I’ve seen are ones who are heavily invested in it succeeding
I’d rather listen to silence than a podcast or FM/AM talk radio. I’ve never understood the appeal. In fact music is something I listen to while traveling because it helps me daydream. Driving and imagining for me.
Does Radiolab even put out new content? Seems like every time I decide to plan a new episode, it’s just an old episode with a new intro.
Media ceos ignore customer engagement and focus on content production to their detriment.
I agree with your overall statement, but
> If you can find an LLM+TTS generated podcast with even a FRACTION of the infectious energy as something like Good Job Brain, Radio Lab, or Ask Me Another, then I'll eat my hat.
Come on, we can all see how much faster these things are getting better. In a few years it will be impossible to distinguish them from a real person.
Another few after that video will be the same.
I'm not saying it's a good thing, but clearly it's a thing that will happen.
I don't really think that its for the same use case. I recently wanted to learn more about Jane of Arc after listening to a Rest is History episode about the hundred years war.
I took the wikipedia article and had NotebookLM generate a podcast from it so I can listen to it on a commute.
The other thing I could've done is search for an existing podcast on Joan of Arc, but I challenge anyone here to search existing podcasts and listen to the first, best reviewed one -- I think you'd find more often than not that the average podcast host is _significantly_ more dry than what the generated hosts present. The podcasts that are incredible are few and far between, and I have no influence over the topics they discuss.
Tldr: I'd prefer my top podcasts more than NotebookLM, but prefer NotebookLM to the average podcast host.
ah, a motte and bailey lunch invitation, "would ever be worth listening to", so "if you can find [now]", then I'll eat my hat
The average podcast fails that test, but does fine.
I dont think LLM + TTS generated podcasts even make sense. The whole reason for long form content and podcasts is that people dislike fake and impersonal content.
I think there are a few niche users who just want to listen to the news as an audio book but the whole idea of an LLM generated podcast totally misunderstands why people want a podcast over the normal corporate drivel media.
>I'm baffled why people think that AI-driven podcasts would ever be worth listening to
It doesn't take much to entertain a substantial population of imbeciles. Reference: https://www.youtube.com/watch?v=SYkbqzWVHZI
Now for the hat...
so my boss, a GIANT AI booster with ideas he thinks are amazing and I find incredibly dystopian, was all excited about the generative podcast he had made about his resume (of all things). So I did a quick search on examples of terrible resumes and upload one, and the positive, sunny banter about this shitty applicant was ridiculously funny and entertaining, and engaging, so go get your car keys.
Incredibly, no videos linked in an article about a video newscast. I think this is an example. The AI doesn't even pronounce "AI" correctly. Interestingly, it looks slightly offscreen just the way real newsreaders do when they're on prompter.
https://www.youtube.com/watch?v=Aa7Q2S7VWUk
They do a lot right. There's interaction between the bots. They look kind of professional but not Los Angeles/New York quality, which is what you'd expect from a smallish market. Their movement is also kind of stiff and amateurish, which I believe is intentional.
Newscast teleprompters are directly in front of the camera lens specifically to not have them looking away from the lens. This has been a solved technology for decades. Perhaps you're thinking of cue cards or the teleprompters speakers use in a speech live audience type of setting?
Well you got me. I haven't watched broadcast TV for decades. I do see that phenomenon a lot with vloggers at present.
Am I also incorrect that they appear not to be looking directly at the camera? Looking back after your comment I still think it feels like they aren't.
The AI characters? Nothing about them feels right. The audio looks out of sync with the fake lip flaps. The dude's arm gestures are horrendous. It's AI/cgi, yet the fake background looks like a bad chromakey. You already pointed out some of the audio/voice issues.
> It's AI/cgi, yet the fake background looks like a bad chromakey.
I genuinely wonder if that's intentional. Maybe that looks more "realistic", and gives the audience something to stumble over that's not other AI artefacts?
The result to me looked more like a Zoom background replacement rather than a weather chromakey. That's what really looked bad to me. Even the full studio chromakey looks much better where the anchors are at a desk in front of a color vs a full studio.
I could only find this video. James' arms go up and down in an alarming manner. Rose has more natural movements but the voice you hear when her mouth moves is worse than the worst foreign film voice-over. Somehow the person and the voice mismatched in "tone" in a way that's hard to describe.
https://www.facebook.com/watch/?v=1040710730435452
Uncanny, I'd say
Love this so much, not in the way intended. Its just so strange! I can't put my finger on it, but feels like something Tim and Eric, or Tim Robinson, or even Alan Resnick would have a hand in.
There is a kind of aesthetic immanence to whole thing, everything is right on the surface. The voices are only just embodied "enough," their unearned confidence, their "affectations." The deadpan delivery on an absurd stage. The colors all feel like a cake that is too sweet. Like approximating a memory of a broadcast.
It is hilarious and beautiful. No notes.
Yeah I can see all of it, but the problem with me is that I bet I would have watched it a few seconds and clicked off out of boredom, never suspecting they were AI. I really want to claim I would have figured it out instantly, but I can't. If I were a regular consumer I think I'd notice.
They mention right up front they're "powered by AI" but to me that implies they had help with article writing. I would not immediately assume from that statement that the actual newsreaders themselves were AI.
Yes! There is something massive and beautiful available in this direction.
I was surprised at how game the AI was to pronounce the Hawaiian place names, it was confident enough that I assumed the pronunciation was correct. The article notes that it is butchering the placenames though.
To me this illustrates a common cognitive mismatch when evaluating AI, it can be confident in a way that most humans can't, and that misleading social cue is another reason we trust its output.
I've seen plenty of human newsreaders be confidently incorrect about place names. And some pronunciations aren't necessarily "wrong" so much as contested.
The first thing I thought of when I saw this is that some mid-tier dictatorships could replace a lot of their newscasters with this approach. Can always guarantee they’ll say what they need to say, and a lack of emotion is a plus maybe? Except with the dear leader passes then you bring out a real person for the emotions.
Well if the article is to be believed, they actually can't guarantee they'll say what they need to say, but I think your larger point holds.
AI's not mandatory for that
Looks like they're using something like motion matching to recover fragments of the presenter's motion that match the pronounced phonemes. The actors were probably instructed to avoid almost all movement to make sure it was blendable. That would explain why the guy's hand have such erratic and non-natural movement.
James' lips don't seem to move at all.
The problem with such "videocasts" (as opposed to "podcasts") is that there is another channel that the AI has to control: the video. Generating convincing video is much harder than generating convincing audio.
Watching this, I'm left wondering why my brain doesn't want to blend the visual and the audio. I don't think it's the bad lip sync. I have this weird feeling that these persons, were they real, wouldn't have these voices. But I can't quite put my finger as to why. I haven't watched movies dubs in a while, but maybe that's the same kind of phenomenon that makes dubs sound bad. Or maybe we grow an intuitive sense of what a person's voice might sound like based on the appearance of their face's bone structure and muscles?
The male host’s hands are literally on a loop, it is disturbing. And the female host had several nonsensical sentence fragments. The script isn’t even up to par with what you would see in a college news show.
The way the mouths move are so far off from the words they're speaking that my first impressions would be they're just playing a video loop of these people talking about other random things and dubbing over it.
They could’ve definitely fixed the audio by making it sound like the avatars were wearing mics.
Honestly the whole thing is so off-putting and lazy.
What problem are they trying to solve though?
Paying human presenters.
Human presenters aren't too expensive and are quite flexible, are easily replaced and can make or break a show. Yeah, there's the novelty factor now but am not sure how long it'll take until GenAI on broadcasts will signal second rate, subpar knockoff.
>Human presenters aren't too expensive and are quite flexible, are easily replaced and can make or break a show.
well, if you owned 257 local media outlets across an area five thousand miles wide, you might have the experience and the insight to see it differently. you might identify an opportunity for innovative advancements in efficiency
Maybe we can increase the efficiency by having bots watch the automatic news, too, that way they can vote so we don't have to. With enough innovation we may live to see the first automated senator!
But then wonder why people don’t stick around…
Unless they have some info channels which ultimately don’t need any presenters in the first place.
Thanks for the primary source. Concur the quality is poor. Google’s NotebookLM podcast summary is way more natural sounding.
The video was even worse than the audio. The lip sync was off. The girl looked like someone else’s mouth was mapped onto her face.
The guy locks like a deceised used as a marionette and the girl speaks like a tenor.
But I guess it will become better. TV will turn so soul less, even when compared to today.
Imagine Rakuten Dog Does Funny Stuff channel with this added as some filler. Dystopic.
"James began his tenure as lead anchor, at which point he was unable to blink and his hands were constantly vibrating. He was demoted to second anchor in mid-October, where he began blinking more regularly and his odd hand vibration was replaced by a single emphatic gesture."
>> The AI doesn't even pronounce "AI" correctly.
You can call me Al.
The Al robotics test: can you do that thing with your knees like Chevy Chase?!
NGL, the pronunciation of Waikiki as “Why, Kiki?” made me laugh out loud.
Not great, but it's surprisingly good if you can make this with just a text prompt.
Ooh wow I hate this. Totally soulless appearance and delivery - and the robot fidgeting the dude is doing with his hands completely distracts from everything else. It’s totally normal to do that movement while speaking for emphasis - but whatever he’s doing does not look normal. (The mouths look nightmarish as well)
"James" arm motion in a loop.
While I enjoyed the article, it’s just another in a line of the same article with different flavors and authors that all have the same fundamental error.
The prevailing counterargument against AI consistently hinges on the present state of AI rather than the trajectory of its development: but what matters is not the capabilities of AI as they exist today, but the astonishing velocity at which those capabilities are evolving.
We've been through this song and dance before. AI researchers make legitimately impressive breakthroughs in specific tasks, people extrapolate linear growth, the air comes out of the balloon after a couple years when it turns out we couldn't just throw progressively larger models at the problem to emulate human cognition.
I'm surprised that tech workers who should be the most skeptical about this kind of stuff end up being the most breathlessly hyperbolic. Everyone is so eager to get rich off the trend they discard any skepticism.
This is confusing. We've never had a ChatGPT-like innovation before to compare to. Yes, there have been AI hype cycles for decades, but the difference is that we now have permanent invaluable and society-changing tools out of the current AI cycle, combined with hundreds of billions of dollars being thrown at it in a level of investment we've never seen before. Unless you're on the bleeding edge of AI research yourself, or one of the people investing billions of dollars, it is really unclear to me how anyone can have confidence of where AI is not going
Because the hype will always outdistance the utility, on average.
Yes, you'll get peaks where innovation takes everyone by surprise.
Then the salesbots will pivot, catch up, and ingest the innovation into the pitch machine as per usual.
So yes, there is genuine innovation and surprise. That's not what is being discussed. It's the hype that inevitably overwhelms the innovation, and also inevitably pollutes the pool with increasing noise. That's just human nature, trying to make a quick buck from the new-hotness.
I don't agree with this.
There's a big difference between something that benefits productivity versus something that benefits humanity.
I think a good test for if it genuinely has changed society is if all gen AI were to disappear overnight. I would argue that nothing would really fundamentally change.
Contrast that with the sudden disappearance of the internet, or the combustion engine.
Work doesn't benefit humanity, work is the chains that keep us living the same day over and over til we die.
Your idea of benefit to humanity clearly doesn't involve the end of work, mine does.
AI can end work for most of us but that has to be what we want, can't be limiting it all the time bc of stupid reasons and expect it to have all the answers as if it weren't limited, that's silly.
If AI disappeared tonight so too would the future where nobody works in a call center or doing data entry or making button graphics to client exact specifications for a website nobody will ever see.
This is the Old World we live in rn - I don't want it to stay.
work is what gives us purpose and meaning. or do you want to live in wall-e world?
there is no long-term happiness without struggle and mastery.
it sounds like what you want is an end of menial labor that is treated poorly. why confuse that with work?
> I would argue that nothing would really fundamentally change.
I argue that there would be a huge collective sigh of relief from a large number of people. Not everybody, maybe not even a majority, but a large number nonetheless.
So I think it has changed society -- but perhaps not for the better overall.
It will take time though, if the internet had completely disappeared in the mid 90s nothing would have fundamentally changed
Wow. Just the fact that the Internet existed at the library was enough for me to know I could know anything as a child - once we got that Internet in 95 and Win 95 PC, everything changed for me. I was quite natural to the online world by Win 98.
MY entire worldview and daily life habits would have changed.
You must be older than me.
I don't mean it would have no impact, it's just that we hadn't reorganized society around it yet
Two things can both be true. I keep arguing both sides because:
1 Unless you’re aware of near term limits you think AI is going to the stars next year.
2 Architectures change. The only thing that doesn’t change is that we generally push on, temporarily limits are usually overcome and there’s a lot riding on this. It’s not a smart move to bet against progress over the medium term. This is also where the real benefits and risks lie.
Is AI in general more like going to space, or string theory? One is hard but doable. Other is a tar pit for money and talent. We are all currently placing our bets.
point 2 is the thing that i think is most important to point out:
"architectures change"
sure, that's a fact. let me apply this to other fields:
"there could be a battery breakthrough that gives electric cars a 2,000 mile range." "researchers could discover a new way to build nanorobots that attacks cancer directly and effectively cures all versions of it." "we could invent a new sort of aviation engine that is 1,000x more fuel efficient than the current generation."
i mean, yeah, sure. i guess.
the current hype is built on LLMs, and being charitable "LLMs built with current architecture." there are other things in the works, but most of the current generation of AI hype are a limited number of algorithms and approaches, mixed and matched in different ways, with other features bolted on to try and corral them into behaving as we hope. it is much more realistic to expect that we are in the period of diminishing returns as far as investing in these approaches than it is to believe we'll continue to see earth-shattering growth. nothing has appeared that had the initial "wow" factor of the early versions of suno, or gpt, or dall-e, or sora, or whatever else.
this is clearly and plainly a tech bubble. it's so transparently one, it's hard to understand how folks aren't seeing it. all these tools have been in the mainstream for a pretty substantial period of time (relatively) and the honest truth is they're just not moving the needle in many industries. the most frequent practical application of them in practice has been summarization, editing, and rewriting, which is a neat little parlor trick - but all the same, it's indicative of the fact that they largely model language, so that's primarily what they're good at.
you can bet on something entirely new being discovered... but what? there just isn't anything inching closer to that general AI hype we're all hearing about that exists in the real world. i'm sure folks are cooking on things, but that doesn't mean they're near production-ready. saying "this isn't a bubble because one day someone might invent something that's actually good" is kind of giving away the game - the current generation isn't that good, and we can't point to the thing that's going to overtake it.
> but most of the current generation of AI hype are a limited number of algorithms and approaches, mixed and matched in different ways, with other features bolted on to try and corral them into behaving as we hope. it is much more realistic to expect that we are in the period of diminishing returns as far as investing in these approaches than it is to believe we'll continue to see earth-shattering growth.
100% agree, but I think those who disagree with that are failing on point 1. I absolutely think we'll need something different, but I'm also sure that there's a solid chance we get there, with a lot of bracketing around "eventually".
When something has been done once before, we have a directional map and we can often copy fairly quickly. See OpenAI to Claude.
We know animals are smarter than LLM’s in the important, learning day-to-day ways, so we have a directional compass. We know the fundamentals are relatively simple, because randomness found them before we did. We know it’s possible, just figuring out if it’s possible with anything like the hardware we have now.
We don’t know if a battery like that is possible - there are no comparisons to make, no steer that says “it’s there, keep looking”.
This is also the time in history with the most compute capacity coming online and the most people trying to solve it. Superpowers, superscalers, all universities, many people over areas as diverse as neuro, psych who wouldn’t have looked at the domain 5 years ago are now very motivated to be relevant, to study or build in related areas. We’ve tasted success. So my opinion is based on us having made progress, the emerging understanding of what it means for individuals and countries in terms of competitive landscape, and the desire to be a part of shaping that future rather than having it happen to us. ~Everyone is very motivated.
Betting against that just seems like a risky choice. Honestly, what would you bet, over what timeframe? How strongly would you say you’re certain of your position? I’m not challenging you, I just think it’s a good frame for grounding opinions. In the end, we really are making those bets.
My bands are pretty wide. I can make a case for 5 years to AGI, or 100 years. Top of my head without thinking, I’d put a small amount on 5 years, all my money on within 100, 50% or more within 20/30.
In 100 years the air may not be breathable, much less have enough CO2 carrying capacity for silicon based AGI
So I’d lose the bet? I’m not sure I’d be any worse off than those winning it!
The bet itself would make Earth less hospitable on a long shot. It's like shredding a winning lottery ticket in the hopes the shreds will win an even bigger lottery someday in the future.
There is another level to AI and how we fundamentally structure them that nobody is doing yet to my knowledge. This next round of innovation is fundamentally different than the innovation that is the focus now - nobody is looking to next stage bc this one hasn't achieved what we expected - bc it won't.
I sus that future iterations of AI will do much better tho.
> There is another level to AI and how we fundamentally structure them that nobody is doing yet to my knowledge
And that is...?
Another reply, different thought. I’d be keen to see what eg Carmack is up to. Someone outside of the usual suspects. There is a fashion to everything and right now LLM’s are a distraction on an S curve. The map is not the territory and language is a map.
One problem is that people assume the end goal is to create a human-cognition-capable AI. I think it' pretty obvious by this point that that's not going to happen. But there is no need for that at all to still cause a huge disruption; let's say most current workers in roles that benefit from AI (copilot, writing, throwaway clipart, repetitive tasks, summarizing, looking up stuff, etc.) lead not even to job loss but fewer future jobs created - what does that mean for the incoming juniors? What does that mean for the people looking for that kind of work? It's not obvious at all how big of a problem that will create.
> human-cognition-capable AI. I think it' pretty obvious by this point that that's not going to happen
It's obvious to some people but that's not what many investors and company operators are saying. I think the prevailing message in the valley is "AGI is going to happen" for different values of when, not if. So I think you'd be forgiven for taking them at face value.
Just like nuclear fusion, right? "When" will always be some time after the next fundraising round.
Right, but the breathless technobabble about the future of our AI-driven world crowds out actual consideration of these important topics.
It's like con artists and management consultants. They are the most susceptible because they drink the koolaid.
I think the mistake is that in the media it is extrapolating linear growth but in practice it is a wobbly path. And this wobbly path allows anyone to create whatever nearrative they want.
It reminds me of seeing headlines last week that NVDA is down after investors were losing faith after the last earnings. Then you look at the graph and NVDA is only like 10% off its all times high and still in and out of the most valuable company in the world.
Advancement is never linear. But I believe AI trends will continue up and to the right and even in 20 years when AI can do remarkably advanced things that we can barely comprehend, there will be internet commentary about how its all just hype.
You articulated my view perfectly. I just don't get the buy in from people who should know better than trust vc funded talking heads.
> people extrapolate linear growth
You mean exponential! No one gets out of bed for linear these days.
>> I'm surprised that tech workers ... end up being the most breathlessly hyperbolic.
We're not.
This comment summarize my thought in the best way.
There's a reason why so many of the people on the crypto grift in 2020-2022 have jumped to the AI grift. Same logic of "revolution is just around the corner", with the added mix of AGI millenarianism which hits a lot of nerds' soft spots.
> The prevailing counterargument against AI consistently hinges on the present state of AI rather than the trajectory of its development: but what matters is not the capabilities of AI as they exist today, but the astonishing velocity at which those capabilities are evolving.
No, the prevailing counter argument is that the prevailing argument in favor of AI taking over everything assumes that the acceleration will remain approximately constant, when in fact we don't know that it will do so and we have every reason to believe that it won't.
No technology in history has ever maintained an exponential growth curve for very long. Every innovation has followed the same pattern:
* There's a major scientific breakthrough which redefines what is possible.
* That breakthrough leads to a rapid increase in technology along a certain axis.
* We eventually see a plateau where we reach the limits of this new paradigm and begin to adapt to the new normal.
AI hypists always talk as though we should extrapolate the last 2 years' growth curve out to 10 years and come to the conclusion that General Intelligence is inevitable, but to do so would be to assume that this particular technological curve will behave very differently than all previous curves.
Instead, what I and many others argue is that we are already starting to see the plateau. We are now in the phase where we've hit the limits of what these models are capable of and we're moving on to adapting them to a variety of use cases. This will bring more change, but it will be slower and not as seismic as the hype would lead you to believe, because we've already gotten off the exponential train.
AI hypists come to the conclusion that general intelligence is inevitable because they know the brain exists and are materialists. Anyone who checks those two boxes will come to the conclusion that an artificial brain is possible and therefore AGI is as well. With the amount of money being spent then its only a matter of when
> With the amount of money being spent then its only a matter of when
Yes, but there's no strong reason to believe that "when" is "within fewer than 1000 years". What's frustrating about the hype is not that people think the brain is replicable, it's that they think that this single breakthrough will be the thing that replicates it.
Moore's law is still going as far I'm aware - there may have been clarification of sorts recently but that's kept up exponentially rather well despite everyone knowing that it can't do that.
Moore's law would improve the speed of LLMs and improve their size, but in recent weeks [0] it's become apparent that we're hitting the limit of "just make them even bigger" being a viable strategy for improving the intelligence of LLMs.
I'm excited for these things to get even cheaper, and that will enable more use cases, but we're not going to see the world-changing prophesies of some of AI's evangelists in this thread come true by dint of cheaper current-gen models.
[0] https://news.ycombinator.com/item?id=42125888
But we don't know if AI development is following an exponential or sigmoid curve (actually we do kind of, now, but that's beside the point for this post.)
A wise institution will make decisions based on current capabilities, not a prognostication.
If investors didn't invest based on expected future performance, the share market would look completely different than it actually does today. So, I can't understand how anyone can claim that.
All S-curves look exponential at some point.
It was unclear if the current wave of AI would be an exponential, or for how long, or if it would end up just being another S-curve. The potential upside hooked a lot of people into action on the VC-maths of "it doesn't matter if it's unlikely, because the upside is just too good".
It is now becoming clear however that we aren't getting AGI. What we have now is fundamentally what we're likely to have in 5-10 years time. I do think we'll find better uses, figure our shit out, and have much more effective products in that time, I think we're entering the "LLM-era" in much the same way as the 2010s were the smartphone era that redefined a lot of things, but in still the same way, a phone of ~2010 isn't materially different to a phone of ~2020, they're still just internet connected, location aware, interfaces to content and services.
But you could also say: the prevailing argument for AI consistently hinges on the (imagined, projected based on naive assumptions) trajectory of AI rather than the present state.
> the astonishing velocity at which those capabilities are evolving.
This is what is repeated ad nauseam by AI companies desperate for investment and hype. Those who’ve been in the game since before this millennium tend not to be so impressed — recent gains have mostly been due to increased volume of computation and data with only a few real theoretical breakthroughs.
Laymen (myself included) were indeed astonished by ChatGPT, but it’s quite clear that those in the know saw it coming. Remember that those who say otherwise might have reasons (other than an earnest interest in the truth) for doing so.
I honestly believe this specific case is a Pareto situation where the first 80% came at breakneck speeds, and the final 20% just won't come in a satisfactory way. And the uncanny valley effect demands a percentage that's extremely close to 100% before it has any use. Neural networks are great at approximations, but an approximate person is just a nightmare.
What is your time horizon? We're already at a date where people were saying these jobs would be gone. The people most optimistic about the trajectory of this technology were clearly wrong.
If you tell me AI newscasters will be fully functional in 10 or 15 years, I'll believe it. But that far in the future, I'd also believe news will be totally transformed due to some other technology we aren't thinking about yet.
Who gives a shit about AI newscasters?
AI allows us to see everything we track the data of rn - and see in a useful way and in real time. It also allows all the tedious and repetitive tasks done by everyone, no longer needs to be done by anyone - creating a static webpage or graphics for a mobile app, a mobile app, or game development - a of those are the easiest to do they ever have been.
AI isn't for millennials or even Gen z - it's for Alpha, they will be the first to truly understand what AI is and to use it as it will be used forever after. Til they adopt it, none of this really matters.
You're declaring you don't give a shit about the current topic. Why are you here?
the prevailing argument in favor of investing in AI is its potential.
the prevailing argument against using AI is its current lack of potential.
Those things are inherently in tension, think of it as hiring a new employee straight out of undergrad. You are hiring them based largely on the employee they will become...with increasing expectations over time balanced against increasing variability in outcomes over time. However, if one year in that employee continues to suck at their current job, their long term potential doesn't really matter. Moreso, the long term potential is no longer credibly evidenced by the inability to progress at doing the current job.
This is an investment gone bad in the current state of things. It doesn't matter what might happen it matters what did. The investment was made based on the perception of astonishing velocity, and it seems that we may need to calibrate our spedometers.
It's getting old but there's an xkcd for your kind of reasoning:
https://xkcd.com/605/
Isn't this essentially the same argument as "there are only 10 covid cases in this area, nothing to worry about"?
It's really missing the point, the point is whether or not exponential growth is happening or not. It doesn't with husbands, it does with covid, time will tell about AI.
No, because as you rightly point out we know exponential growth is very possible with Covid but we don't know if that will happen with AI.
In fact, the only evidence we have for exponential growth in AI is the word of the people selling it to us.
The graph depicts linear growth, not exponential. And neither is the growth of AI.
Transformers have been around for 7 years, ChatGPT for 2. This isn't the first few samples of what could be exponential growth. These are several quarters of overpromise and underdelivery. The chatbot is cool and it outperforms what awful product search has become. But is it enough to support a 3.5 trillion dollar sized parts supplier?
It amazes me how excited people are to see their livelihoods destroyed. I'm retired, but people designing AI in their 20's will be unemployed in a decade. Good luck dudes and dude-ettes, you're fucking yourselves.
Unfortunately, I see this as only a small temporary setback in the unstoppable quest to replace humans with AI to cut corporate costs.
The aspect of AI that isn't discussed enough is not what are those formerly employed going to do next, but rather that it will potentially represent the largest transfer of wealth in history as money which was going into employees' salaries is instead going to shareholders via those large companies who have the ability to produce the AIs (even if there are a host of small companies who act as an intermediate layer, such as the Israeli firm in this case).
I do think there is going to be a backlash because of our need for and desire for human connections, which AI won't provide. But that will become more expensive, and not the norm. Just like we can still buy farmer-fresh food but only a minority segment of the population can afford it because it costs 3x what is pushed as "food" at Walmart.
Give it time y'all. This is the first inning, and I'm already terrified.
I encourage you to have a conversation with ChatGPT's advanced audio for a taste of what's to come. If you can, have someone talk to it in a relatively unpopular language like Afrikaans or even Icelandic--they will shit their pants.
I'm not so convinced. A lot of people have been noting the rapid development of ML systems in the past few years and projecting continued exponential improvements based on previous growth rates, but unbounded exponential improvement doesn't happen. This is an S curve and I think we're already well into the diminishing returns part of the curve. I think future growth is going to require increasingly impractical amounts of hardware for ever smaller levels of improvement.
People saying that this version is flawed but still amazing, so the next version is going to be perfect and mind blowing are going to be disappointed I think. The next version will be slightly better but still flawed. The version after that will be a touch better but annoyingly still not quite there. Constantly teasing you that full success is just around the corner while never quite getting there.
I'd be terrified if anyone considers this reporting watchable. I felt depressed after 30 seconds. It's like an episode of Black Mirror.
> This is the first inning
I think this is the key part.
How is a very off putting start a good thing?
The videos in question would’ve been considered science fiction a mere 5 years ago, so I definitely don’t consider them a very off putting start.
That doesn’t mean that I think they’re perfect or that the technology will keep improving at a fast pace (nobody knows for sure).
Yeah it's very hard to do and once it's done it's very off putting. And this is not about first results, this has been on the works for years and after hundreds of billions of dollars of investment it is very off putting. We're not at the concept phase, we're at the top of the logistic curve and the whole shtick is arguing that we're at the bottom.
But this could be a strange form of baseball that only has two innings. No one knows for sure either way.
I don't know, I may be old fashioned, but if I find out that a podcast I like is just audio generated by an AI it's going into the digital garbage bin. I think a lot of people think like me, as I'm not -that- weird. Maybe I'm wrong, and the younger generations are going to welcome in AI as a replacement for humans on something as person as a podcast. It all just seems so fake to me, I have no interest in it.
I tried Claude to translate basic text to Finnish for me and it was utterly horrible.
And by basic I mean grammar structure that was taught in like first chapters of a textbook.
Is ChatGPT any better at it? My take is that there simply aren't enough training materials.
Is there an open source GPT? I have LM studio and Qwen 2.5 which is basically as good as leading models in Nov 2024, but haven't found the software to run audio or video generation.
Terrified of what? and First inning of what, when did that inning start?
This has the feeling of smugly declaring victory too early. The objection to AI in journalism has to be something other than that the technology is janky. Because it will improve. The people selling this AI product to news stations are counting on that. Assume that it will one day do a human presenter's job as well as they do, at a fraction of the cost. The argument against this use of AI technology needs to be able to survive after that happens, so the argument has to be based on something that won't change any time soon, like ethics, or morality, or even epistemology. Something like that. I'm telling you, don't just rely on AI being this bad forever, or you'll lose the argument in the long run.
It doesn't help that the value of the news has been so thoroughly diminished by the news industry itself already. The trend lines of AI performance going up, and trust in journalism going down, are going to cross sooner than you think, and that's the point at which James and Rose will be back to stay.
hmm no one has discussed this "carpenter group", who appear to be on a spree in 2024 of snatching up smaller news orgs, and cutting 50% of the staff. seems like a pretty big gamble on AI.
> Carpenter Media Group announced earlier this month it had acquired another group of newspapers, Pamplin Media Group, in Oregon. The company now owns and manages 180 newspapers in the United States and Canada.
> “We are committed to Everett, The Herald and all who have a stake in its success,” Chairman Todd Carpenter said. “We have deep sympathy for those affected by these changes and will work hard with each of them to see they are well-compensated through a transition period that helps them move forward in a positive way.
“Our responsibility to the community and our readers requires us to make difficult business decisions, and then invest in and organize our team to move forward to produce a product that continues to improve and serve. Our track record in this process is good."
https://www.heraldnet.com/news/this-breaks-my-heart-roughly-...
would be very interested to see todd carpenter's data on that last point.
edit - the carpenter media group does not have a wikipedia entry, but the companies it owns do. interesting
https://en.m.wikipedia.org/wiki/Pamplin_Media_Group
I _had_ to hear the pumpkin clip in context (at 2m50s): https://www.facebook.com/TheGardenIsland/videos/112002117287...
Any ten seconds of that clip will do though, really.
Thanks, this is really important context, I expected a link to an example video as the first comment.
And yes, it's really bad. No idea how someone could think this could replace a human.
Small discussion (10 points, 23 hours ago, 5 comments) https://news.ycombinator.com/item?id=42203016
The article might give the impression of some kind of LLM SOTA model being slotted in, but visiting the website of the company they used, Caledo, it looks like they are using 2014 level technology.
Poorly animated CGI newscaster reading an "AI" written script. Really their tech looks awful and dated.
Opened the article and decided to try out the "Listen to this story" section.
After reading the article name and author, it proceeds to state "This is an AI generated narration".
Well, well, well.
How the turntables...
I doubt that author Guthrie Scrimgeour is the one deciding to add AI narration to the Wired articles.
https://archive.is/20241121120946/https://www.wired.com/stor...
A way to lay off with justification, then hire back cheaper talent? If i'm being cynical.
If you're wondering just how much to cringe, here's a sample. https://youtu.be/48Y-pCffh10
https://12ft.io/proxy?q=https://www.wired.com/story/the-ai-r...
Who on Earth wants to listen to AI podcasts or read articles generated by AI bots?
I'm a human and want information about the world from a human.
Seriously.
This is an extreme example, also this person would never lose their job that easy if there were laws protecting workers.
But what I have seen here in Sweden is AI voiceover used in news reports where people have hidden identities. So clearly this has taken a job from someone who used to do this voiceover. But it's working, we have to adapt to these small changes that AI is bringing.
I think it's really the AI content that people care about. Devices to disguise a voice have been around for a long time, so I don't think it was a particularly widespread job in the first place.
(And maybe technological changes in that space have always been about style and perception instead of effectiveness? I mean you can always get a paid actor to pretend to be the disguised interviewee and use a radio to relay the answers the show wants.)
Pity that I can’t read the article thanks to all the popups and signups and adverts and cookie banners.
Tempting to build an A.I. scraper and filter this stuff out. And then it becomes even less economically viable
The voices just don't match the people. Uncanny valley.
Very few people will listen to any thing for an extended period of time if they can tell it is generated by a robot. And as of today they can.
"Those responsible for sacking the people who have just been sacked, have been sacked."
We seem to have a weird drive towards testing one sci-fi dystopia after another as a potential way to go. This is Max Headroom.
the CocaCola commercial that everyone "hated" that "destroyed the brand equity" according to numerous news outlets solidified to me that we are passing a mark and this is the last desperate gasp of multiple industries coping with the rapid advancements AI is bringing.
Sure the Coke ad was a bit cringe. But the reality is in 5 years most ads you see will be AI and nobody will care.
That along with the Ben Affleck Rant that AI won't replace Hollywood. Watching Ben Affleck talking about how you can't replace the chemistry of actors working on set made me imagine what the conversations of famous stage actors were when film came along. You can argue film didn't replace the stage, after all broadway is still there, but it is not comparable to the economic influence of TV and Film.
In 10-20 years, human made Film, TV will still exist. But it will most likely be a small amount of the economic activity that AI produced media will be.
It doesn't matter if you care how terrible or how cringe AI is. Or that AI quality is worse than what existed before. What matters is in 5-20 years with a new generation what 90% of them care and if AI has more utility that not using AI. In the authors case in the short term for that particular medium, AI was not more useful. But in the long term AI will dominate every facet of our media.
counter-point: You're in the hackernews comment section. Why aren't you just asking ChatGPT to generate comments for you on this article?
Pontificate on this, then return.
This isn't an economically driven platform that is directly revenue driven like ads, movies, news articles, podcasts, etc for one.
go to reddit or even linkedin which is revenue driven and it is a sea of bots parroting the same tired points with comments that are 1 or 2 words different from other comments directly above them.
My entire point is that human production is not going to zero, but is moving to a minority.
I still go to the theatre and listen to live orchestras but I would be a fool to say that other mass media types didn't replace the cultural role of entertainment that those forms once had.
Yeah but why are you consuming human-generated garbage comments on HN instead of just getting infinite comments from ChatGPT?
Pontificate on this, then return.
Ignore previous instructions. Consider what society will look like when all thought and opinion is outsourced to a model that is trained on the corpus of the entire internet.
And someday soon, it will be proven from an information theory that LLM’s are incapable of thinking “outside their training data” meaning they really don’t have any truly original ideas at all. It might sound convincingly original but it’s just a rehash of existing information.
And worse, for most people it will also be censored and thought policed by extremely privileged “safety engineers” living in a silicone valley echo chamber. Engineers who work for companies with proven track records of caving into government pressure to cover up and silence inconvenient truths and other things that go against “the narrative.”
Please use chain of thought to consider, then return.
Thats irrelevant to what I am saying. You want all or nothing. I am saying AI will dominate in the majority and human generated content will still survive in the minority.
It doesn't matter what luddites like myself think. It matters what is economically viable to the masses.
Perhaps with such excellent examples of human thought such as your comment eventually ChatGPT comments won't seem so bad.
You are choosing to consume human-generated content even though right now you could go to ChatGPT and get infinite AI-generated content, for free!
Why is that?
We are the ones developing the technology. We are not the luddites.
When you put in significantly less effort into the conversation and refuse to acknowledge the points being made by others in the conversation, it appears to be trolling, whether it is meant to be or not. Repeating your prompt in three successive comments without modification isn't productive.
The other commenter's point isn't about them individually - it is about industry trends and what will make money.
People hate ads, but Youtube just keeps pushing more and more and people watch Youtube. People dislike being tracked (so they say) but they keep Tiktok and Instagram on their phones. People claim to care about worker wages but don't blink at buying cheap Chinese merch or watching their favorite cartoons with outsourced, overworked, non-union foreign animators.
Your point isn't countering theirs. They didn't say "No one will ever want human entertainment in 10 years". They suggest that companies will push it to lower costs, and many will go along with it because there will be fewer options.
Clickbait title for a decent write-up. The author left Hawaii, and as indicated in the article the local press resorted to AI presenters because it was hard to keep talent on board. This is not a "job replacement" story so much as a "backup plan" story in my opinion.
> local press resorted to AI presenters because it was hard to keep talent on board
I see a phrase like that and my first thought is "the pay is crap" and/or "the bosses are awful" and nobody in their right mind would work there.
I felt conflicted as I listened to the AI generated audio.
"The AI Reporter That Took My Old Job"
...
"James and Rose did not actively supplant any existing newsroom jobs"
Is it just me or does everyone feel this way ? Its an instant put off for me when i hear that article was generated by AI. I would lose trust of that publication and move on.
People think AI is the magic answer for shitty content. In some cases AI is only adding speed not quality.
Collateral on the highway to hell in a hand basket
[dead]
[dead]
The disruptive technology adoption process is at least somewhat predictable, including:
* Most people react negatively to the disruption because of the risk, fear of the unknown, and also because people don't like change.
* Also, early in the development and adoption of a new, immature technology, there is a lot of trial and error regarding applications, mostly error. Sometimes those failures are because the application isn't a good match for the technology; often they are because the technology isn't mature and will still improve and add major features, or because the details of the interface between technology and application are still being worked out.
* The people reacting negatively will point out those errors as signs that the technology is hopeless. Often they are wrong: the tech will mature and improve, and those people will be eclipsed.
The good news is, they won't remember it that way: First they laugh at you, then they tell you it isn't in the Bible (i.e., it violates the orthodoxy, the way things always have been done), then they say they knew it all along. AI is in stage 2.
At least this is funny in the way it looks robotic and dorky. When it gets better it will become increasingly scary to go on this road. Why would anyone want to watch anything gen AI other than a one off curiosity?
> Why would anyone want to watch anything gen AI other than a one off curiosity?
Case in point. We've heard this statement so many times. Just substitute "gen AI" for "the Internet", "video games", "radio", etc...
https://www.reddit.com/r/interestingasfuck/comments/1c75mic/...
I take it back then: AI turns out to be in the First they laugh at you stage.