Most AI layoffs will be white collar jobs. Voting/party blocs that have large blue collar bases won't care and have already been dealing with similar pressues, e.g. NAFTA and such, for decades. So there's going to be a strong sense of "haha now it's happening to you like it happened to us in the 80s and 90s." It won't be so bad. Cities may become cheap and fun again.
"... they could pass laws establishing and funding retraining programs that teach workers how to work alongside AI systems"
Computers were billed as the entryway to the future to the masses in the 90's and 2000's. So we got things like "send everyone to college" and ITT Tech. That didn't work, but a lot of money was wasted and a few people made a lot of money. This will be a repeat.
If a local area is seeing blue collar industry as the future, it needs cheap labor, which needs cheap housing near its plants. Therefore housing near industrial sections of cities needs to get cheaper. Converting tall office builings to dense apartments could happen. Especially if blue collar industries simply buy the empty buildings, assuming they aren't converted to housing for GPUs first.
I too believe social unrest due to labor market disturbances is very likely. But I will feel some relief if that's all it comes from it. AI has the potential to automate everything, including company management and civil governance, at which point people will be out of the loop to a degree not seen before in history. The worst part is that it won't happen overnight; society will gradually anneal humans towards that redundancy potentially without any shock that would provoke our species to react decisively. Plus, we are starting from a bad place already.
AI does not have that potential. Not now, maybe in some distant future, as in decades from now.
Right now, AI has the potential to increase productivity. Work that's done by 10 people without AI, can be done by 5 people with AI. So 5 people can be laid off. But soon, companies will realize that rather than laying off half of their workforce, they can simply produce twice as much as before.
In reality, the AI productivity boost is nothing close to 100%. Maybe 10% per year. That could translate in 5% layoffs, 5% increased output. Nothing extraordinary compared to other times in history.
Even following your model, in twenty years we end up with 0.95*20=0.36, that's a whopping 64% of the population not-needed. People will do their best to adapt, but the article rightly points out that even when blue-collar jobs were off-shored and there were plenty of white-collar jobs, most people couldn't adjust in a decade or two, and that it still resulted in significant social strain.
Now imagine most white-collar jobs gone too, our cracked democratic institutions finally crumbling, and survtech and repressive police everywhere to squash protesters and dissidents... I've coined a term for a way to get out of that dystopia, if we manage: The Grim Revolts[^1]. I hope it remains the fiction of my feverish mind that it's supposed to be, and that we do manage to find good ways to prevent things from spiraling.
> Right now, AI has the potential to increase productivity.
The detail many seem to forget is that "AI" is Artificially Inexpensive. It must eventually turn a profit, or collapse. Once providers start charging what they must to remain solvent, and/or the output is burdened with advertising, the gains, real or perceived, will likely evaporate.
I disagree. AI is very cheap. People speculate that OpenAI and Anthropic and Google heavily subsidize the AI they provide. But all the evidence points towards this not being true. You can look at all the independent providers, like Cerebras, Groq, TogetherAI, and dozens others. Some may swim in venture capital money, and can afford to subsidize, but I doubt all can do that. And if they can't, then how do you explain that the cost of the million tokens is so low?
And this is just now. Inference costs are plummeting, because models are becoming more and more efficient. I can get 6 tokens/second on my local Ollama from GPT-OSS-20B using only CPU, and I can get 11 tps from Qwen3-30B. This was unthinkable 6 months ago. I am quite certain I'll get faster speeds 6 months from now and faster still 6 months later. Model architectures are becoming better and better, and models with the same number of parameters are becoming smarter and smarter.
If AI is wildly successful and improves significantly beyond current capabilities and takes all of the jobs, the most aggressive adopters will self-terminate. The employers that survive this will not be those who can cut costs the most, but those who are most critical (and that will tend to be something that interfaces with the physical world, less digital). If it comes down to fixing the plumbing leak or Netflix, I guess I’ll be less entertained, or go outside. No amount of cost reduction survives zero revenue when no one has an income.
If AI continues at its current trajectory of being an incrementally improving blog post generator, then the only losers here are the big investors.
To see this, I would be happy in so many contexts and situations. I'm not by nature a luddite and appreciate the value of AI in many things, even as it now is, but the sheer overload of shoehorning it into every possible thing for better KPIs while maximizing customer service "fuck-them value" by idiot executives and moron managers (choke on your own feces, all of you) still hoping to latch their suckers on the STILL BIG thing of the FUTURE!!! without a clue about its details and limits is revolting. Enshittification has been thus exponentiated!
Then, there's seeing government use it in its own disgustingly parasitical ways for accelerated mass surveillance, and to not even mention the absolute flood of utterly brain-dead spam sludge content that has flooded all social networks, online searches and just about every corner of the internet (including my inbox, thanks mom and friends who have no clue how to distinguish Ai slop from anything in the least bit valuable)
And I just want a large part of it to burn and die back. I'm with the artists and real photographers on this one.
Our modern view of the Luddites is not a fair shake. The Luddites were not opposed to technology in general, they merely (and understandably) opposed having their own livelihoods replaced by automation.[1] It's a familiar sentiment in these times.
Incidentally, I'm not implying that you don't know what the original Luddites were about; I add my comment for the benefit of anyone else who may be unfamiliar with the backstory.
One clear way that economic theory (and citizen preference) doesn’t match government action is that when change causes job losses the correct answer is job retraining.
And yet job retraining programs do not exist (that I know of) and certainly not at any scale comparable to the job losses.
Most AI layoffs will be white collar jobs. Voting/party blocs that have large blue collar bases won't care and have already been dealing with similar pressues, e.g. NAFTA and such, for decades. So there's going to be a strong sense of "haha now it's happening to you like it happened to us in the 80s and 90s." It won't be so bad. Cities may become cheap and fun again.
"... they could pass laws establishing and funding retraining programs that teach workers how to work alongside AI systems"
Computers were billed as the entryway to the future to the masses in the 90's and 2000's. So we got things like "send everyone to college" and ITT Tech. That didn't work, but a lot of money was wasted and a few people made a lot of money. This will be a repeat.
Cities aren’t expensive because of white collar jobs.
If a local area is seeing blue collar industry as the future, it needs cheap labor, which needs cheap housing near its plants. Therefore housing near industrial sections of cities needs to get cheaper. Converting tall office builings to dense apartments could happen. Especially if blue collar industries simply buy the empty buildings, assuming they aren't converted to housing for GPUs first.
lol this article has a Motte and Bailey where it declares that a country favoring domestic trade over foreign trade is now "anti-trade"
I too believe social unrest due to labor market disturbances is very likely. But I will feel some relief if that's all it comes from it. AI has the potential to automate everything, including company management and civil governance, at which point people will be out of the loop to a degree not seen before in history. The worst part is that it won't happen overnight; society will gradually anneal humans towards that redundancy potentially without any shock that would provoke our species to react decisively. Plus, we are starting from a bad place already.
AI does not have that potential. Not now, maybe in some distant future, as in decades from now.
Right now, AI has the potential to increase productivity. Work that's done by 10 people without AI, can be done by 5 people with AI. So 5 people can be laid off. But soon, companies will realize that rather than laying off half of their workforce, they can simply produce twice as much as before.
In reality, the AI productivity boost is nothing close to 100%. Maybe 10% per year. That could translate in 5% layoffs, 5% increased output. Nothing extraordinary compared to other times in history.
Even following your model, in twenty years we end up with 0.95*20=0.36, that's a whopping 64% of the population not-needed. People will do their best to adapt, but the article rightly points out that even when blue-collar jobs were off-shored and there were plenty of white-collar jobs, most people couldn't adjust in a decade or two, and that it still resulted in significant social strain.
Now imagine most white-collar jobs gone too, our cracked democratic institutions finally crumbling, and survtech and repressive police everywhere to squash protesters and dissidents... I've coined a term for a way to get out of that dystopia, if we manage: The Grim Revolts[^1]. I hope it remains the fiction of my feverish mind that it's supposed to be, and that we do manage to find good ways to prevent things from spiraling.
[^1]: https://w.ouzu.im/
> Right now, AI has the potential to increase productivity.
The detail many seem to forget is that "AI" is Artificially Inexpensive. It must eventually turn a profit, or collapse. Once providers start charging what they must to remain solvent, and/or the output is burdened with advertising, the gains, real or perceived, will likely evaporate.
I disagree. AI is very cheap. People speculate that OpenAI and Anthropic and Google heavily subsidize the AI they provide. But all the evidence points towards this not being true. You can look at all the independent providers, like Cerebras, Groq, TogetherAI, and dozens others. Some may swim in venture capital money, and can afford to subsidize, but I doubt all can do that. And if they can't, then how do you explain that the cost of the million tokens is so low?
And this is just now. Inference costs are plummeting, because models are becoming more and more efficient. I can get 6 tokens/second on my local Ollama from GPT-OSS-20B using only CPU, and I can get 11 tps from Qwen3-30B. This was unthinkable 6 months ago. I am quite certain I'll get faster speeds 6 months from now and faster still 6 months later. Model architectures are becoming better and better, and models with the same number of parameters are becoming smarter and smarter.
If AI is wildly successful and improves significantly beyond current capabilities and takes all of the jobs, the most aggressive adopters will self-terminate. The employers that survive this will not be those who can cut costs the most, but those who are most critical (and that will tend to be something that interfaces with the physical world, less digital). If it comes down to fixing the plumbing leak or Netflix, I guess I’ll be less entertained, or go outside. No amount of cost reduction survives zero revenue when no one has an income.
If AI continues at its current trajectory of being an incrementally improving blog post generator, then the only losers here are the big investors.
To see this, I would be happy in so many contexts and situations. I'm not by nature a luddite and appreciate the value of AI in many things, even as it now is, but the sheer overload of shoehorning it into every possible thing for better KPIs while maximizing customer service "fuck-them value" by idiot executives and moron managers (choke on your own feces, all of you) still hoping to latch their suckers on the STILL BIG thing of the FUTURE!!! without a clue about its details and limits is revolting. Enshittification has been thus exponentiated!
Then, there's seeing government use it in its own disgustingly parasitical ways for accelerated mass surveillance, and to not even mention the absolute flood of utterly brain-dead spam sludge content that has flooded all social networks, online searches and just about every corner of the internet (including my inbox, thanks mom and friends who have no clue how to distinguish Ai slop from anything in the least bit valuable)
And I just want a large part of it to burn and die back. I'm with the artists and real photographers on this one.
Rant over.
> I'm not by nature a luddite
Our modern view of the Luddites is not a fair shake. The Luddites were not opposed to technology in general, they merely (and understandably) opposed having their own livelihoods replaced by automation.[1] It's a familiar sentiment in these times.
Incidentally, I'm not implying that you don't know what the original Luddites were about; I add my comment for the benefit of anyone else who may be unfamiliar with the backstory.
[1] https://en.wikipedia.org/wiki/Luddite
Not to mention execs lazily scapegoating AI for every layoff and firing, even though AI has little to do with it
One clear way that economic theory (and citizen preference) doesn’t match government action is that when change causes job losses the correct answer is job retraining. And yet job retraining programs do not exist (that I know of) and certainly not at any scale comparable to the job losses.