I don't personally think there's that much value to this argument. Compare, for instance the consumption of 1 hour of tv vs. 1 hour of GPT usage:
A single AI chat message can consume 0.34 watt-hours of energy (1). So, let's say a hundred messages in an hour (quite an aggressive session) would be 34 watt-hours of energy.
An LCD TV running for an hour consumes about 100 watt-hours of energy, depending on size, LED, vs. OLED etc. (2).
I think AI does help people do better research faster, which is a significant uplift to humanity, while I do not see anyone specifically curbing their TV usage. We should probably focus our effrots on helping people use AI better and meanwhile build more nuclear energy plants, imo.
> I think AI does help people do better research faster, which is a significant uplift to humanity,
One must consider both sides. Better research also contributes (no question) to more consumerism and the furthering of technology, which also uses more fossil fuels. It's time we acknowledged that research isn't free, and all of the damage to the biosphere was mainly enabled by science.
And AI uses a huge ton of energy. For example, according to [1], "In Ireland, [...] electricity demand from data centres represented 17% of the country’s total electricity consumption for 2022". And we also have to consider the raw materials and mining used.
1. In Ireland, where the data centre market is developing rapidly, electricity demand from data centres represented 17% of the country’s total electricity consumption for 2022
AI is psychologically dangerous. It's the greater risk. It can influence people do to things.
The environment talk is perhaps relevant, but if eventually AI becomes ecologically friendly, then people who are against it would be left with no argument.
I'm skipping over to the core of the problem right away. It's a risk to human minds, always. The eco talk is probably relevant, but most likely it's more resounding as a bait to make you look like a tree-hugger that doesn't know what you're talking about.
To an extent, yes, but companies are investing more into renewable sources of energy anyway, and you might even get a demand-side effect where the cost of supply of energy is incentivized to go down due to higher competition.
But I assume your friends are justifying this in more of an ideological way than you'd think, because there are lots of things that are more environmentally damaging than data centers, and even for data centers it's not like we haven't been using Google and Facebook for 20 years when the same concerns of energy would've been there.
What I've seen is people dressing up the Luddite argument of losing economics of living into a lot of other arguments like energy usage, when in reality I doubt they really give a shit about that over environmental degradation in other areas of their life.
RAM is the reason LLMs are so power inefficient. Shuttling weights and results from RAM to compute and back for everything is where most of the power goes.
It doesn't have to be that way. For a sufficiently large load, it makes sense to use reconfigurable hardware and bake in the constants and s dataflow at runtime.
Think of it like using an array of FPGAs large enough to hold the whole model unwound, yet that could be configured in seconds at runtime. You'd get tokens at 100 MHz or more .
You would think saving 95% or more on power and infrastructure for a given token rate would be worth it, especially when contemplating Trillion dollar outlays.
Many things don't have to be the way they are. But as long as the powerful big tech can subsidize their costs on the commons of the environment in the form of environmental damage without regulation, they will only pay lip service to making things more efficient. Money is a much more powerful motivator to the unscrupulous than protecting the long-term health of the commons.
Not if spending a little extra money and keeping with the inefficiency helps make money in other ways, such as getting the product out faster or allowing their workers to focus on the tech stack. Saving a little electricity cost might cost them in their development, so they're likely to use the cheap electricity and offset the cost to the environment.
Of course you are wrong. "IEA's models project that data centres will use 945 terawatt-hours (TWh) in 2030, roughly equivalent to the current annual electricity consumption of Japan." [1] And don't forget the recent announcement by Fermi America to build a 6 gigawatt nuclear datacenter in Texas. And what are we getting in return:
(1) AI is primarily used, by and far, to accelerate consumerism.
(2) AI has very few applications that are actually solving the world's problems (the world's problems are mostly nontechnical problems). The cited ones of scientific or medical are either too abstract, or else they are likely to make the problem worse. And in the case of medical, maybe the AI applications will help a few hundreds of thousands of people – but hurt/kill many more with its contribution to climate change. So not worth it.
I consider those who promote AI to be enemies of humanity and biological life since they are using a commodity that we should be using less of with reckless abandon.
Nuclear is good though, it's the single densest source of energy in the world (and, well, outside of it too).
I don't get your argument though, where has it "accelerated consumerism?" And you can't be serious about its medical applications, I doubt more people would die from its incremental climate change effects (as compared to, say, car accidents, much less car pollution in cities) as compared to being saved from death. Even outside of the scientific fields, AI sure is solving a lot of my problems, saying one is an "enemy of humanity" is the sort of hyperbole I'd only see on HN.
> Nuclear is good though, it's the single densest source of energy in the world (and, well, outside of it too).
It would be good if the energy were used in critical applications. Wasted energy is stil wasted.
> I don't get your argument though, where has it "accelerated consumerism?"
Are you serious? It is making people richer, allowing people to make products faster, including software. Of course that makes consumption faster.
> And you can't be serious about its medical applications, I doubt more people would die from its incremental climate change effects (as compared to, say, car accidents, much less car pollution in cities) as compared to being saved from death.
No, I am serious. And it will get much worse in the future. Check out [1], [2], etc.. And lets not forget the increased storms, flooding, etc. We are all responsible for that. But those that use electricity like big tech should be held especially responsible because they encourage the bad behaviour and use resources directly.
> Even outside of the scientific fields, AI sure is solving a lot of my problems, saying one is an "enemy of humanity" is the sort of hyperbole I'd only see on HN.
Solving your problems doesn't really mean solving the world's problems, just like making the top 10% richer doesn't make the world a better place. I absolutely consider them the enemy of humanity.
>I keep getting told by friends not to use AI because it is destroying the environment.
I side with them. If you're environmentally conscious you absolutely cant support AI. The amount of energy and the backing of basically every AI is fossil fuels.
Let's not also forget material use. According to [1], "Global AI demand is expected to consume 4.2-6.6 billion cubic meters of water by 2027, surpassing Denmark’s total annual water withdrawal of 4-6 billion cubic meters." And there's also all the mining required for the materials to build the computers, and the fossil fuels shipping them and making them in the first place in places like Taiwan.
I don't personally think there's that much value to this argument. Compare, for instance the consumption of 1 hour of tv vs. 1 hour of GPT usage:
A single AI chat message can consume 0.34 watt-hours of energy (1). So, let's say a hundred messages in an hour (quite an aggressive session) would be 34 watt-hours of energy.
An LCD TV running for an hour consumes about 100 watt-hours of energy, depending on size, LED, vs. OLED etc. (2).
I think AI does help people do better research faster, which is a significant uplift to humanity, while I do not see anyone specifically curbing their TV usage. We should probably focus our effrots on helping people use AI better and meanwhile build more nuclear energy plants, imo.
(1): https://epoch.ai/gradient-updates/how-much-energy-does-chatg...
(2): https://santannaenergyservices.com/how-many-watts-does-a-tv-...
---
And then consider the amount of energy traditionally required by one human to do the same research tasks. Also quite significant.
I think we should be focused on making the more efficient, for sure! But I don't buy that the arguments based on energy consumption are very strong.
> I think AI does help people do better research faster, which is a significant uplift to humanity,
One must consider both sides. Better research also contributes (no question) to more consumerism and the furthering of technology, which also uses more fossil fuels. It's time we acknowledged that research isn't free, and all of the damage to the biosphere was mainly enabled by science.
And AI uses a huge ton of energy. For example, according to [1], "In Ireland, [...] electricity demand from data centres represented 17% of the country’s total electricity consumption for 2022". And we also have to consider the raw materials and mining used.
1. In Ireland, where the data centre market is developing rapidly, electricity demand from data centres represented 17% of the country’s total electricity consumption for 2022
AI is psychologically dangerous. It's the greater risk. It can influence people do to things.
The environment talk is perhaps relevant, but if eventually AI becomes ecologically friendly, then people who are against it would be left with no argument.
I'm skipping over to the core of the problem right away. It's a risk to human minds, always. The eco talk is probably relevant, but most likely it's more resounding as a bait to make you look like a tree-hugger that doesn't know what you're talking about.
To an extent, yes, but companies are investing more into renewable sources of energy anyway, and you might even get a demand-side effect where the cost of supply of energy is incentivized to go down due to higher competition.
But I assume your friends are justifying this in more of an ideological way than you'd think, because there are lots of things that are more environmentally damaging than data centers, and even for data centers it's not like we haven't been using Google and Facebook for 20 years when the same concerns of energy would've been there.
What I've seen is people dressing up the Luddite argument of losing economics of living into a lot of other arguments like energy usage, when in reality I doubt they really give a shit about that over environmental degradation in other areas of their life.
RAM is the reason LLMs are so power inefficient. Shuttling weights and results from RAM to compute and back for everything is where most of the power goes.
It doesn't have to be that way. For a sufficiently large load, it makes sense to use reconfigurable hardware and bake in the constants and s dataflow at runtime.
Think of it like using an array of FPGAs large enough to hold the whole model unwound, yet that could be configured in seconds at runtime. You'd get tokens at 100 MHz or more .
You would think saving 95% or more on power and infrastructure for a given token rate would be worth it, especially when contemplating Trillion dollar outlays.
Many things don't have to be the way they are. But as long as the powerful big tech can subsidize their costs on the commons of the environment in the form of environmental damage without regulation, they will only pay lip service to making things more efficient. Money is a much more powerful motivator to the unscrupulous than protecting the long-term health of the commons.
It costs money to run a datacenter, so even if you consider money to be their motivator, it benefits them to make them more efficient.
Not if spending a little extra money and keeping with the inefficiency helps make money in other ways, such as getting the product out faster or allowing their workers to focus on the tech stack. Saving a little electricity cost might cost them in their development, so they're likely to use the cheap electricity and offset the cost to the environment.
The people who work on the product are not also working on the data centers
Of course you are wrong. "IEA's models project that data centres will use 945 terawatt-hours (TWh) in 2030, roughly equivalent to the current annual electricity consumption of Japan." [1] And don't forget the recent announcement by Fermi America to build a 6 gigawatt nuclear datacenter in Texas. And what are we getting in return:
(1) AI is primarily used, by and far, to accelerate consumerism.
(2) AI has very few applications that are actually solving the world's problems (the world's problems are mostly nontechnical problems). The cited ones of scientific or medical are either too abstract, or else they are likely to make the problem worse. And in the case of medical, maybe the AI applications will help a few hundreds of thousands of people – but hurt/kill many more with its contribution to climate change. So not worth it.
I consider those who promote AI to be enemies of humanity and biological life since they are using a commodity that we should be using less of with reckless abandon.
1. https://hardware.slashdot.org/story/25/04/10/2019233/data-ce...
> nuclear datacenter
Nuclear is good though, it's the single densest source of energy in the world (and, well, outside of it too).
I don't get your argument though, where has it "accelerated consumerism?" And you can't be serious about its medical applications, I doubt more people would die from its incremental climate change effects (as compared to, say, car accidents, much less car pollution in cities) as compared to being saved from death. Even outside of the scientific fields, AI sure is solving a lot of my problems, saying one is an "enemy of humanity" is the sort of hyperbole I'd only see on HN.
> Nuclear is good though, it's the single densest source of energy in the world (and, well, outside of it too).
It would be good if the energy were used in critical applications. Wasted energy is stil wasted.
> I don't get your argument though, where has it "accelerated consumerism?"
Are you serious? It is making people richer, allowing people to make products faster, including software. Of course that makes consumption faster.
> And you can't be serious about its medical applications, I doubt more people would die from its incremental climate change effects (as compared to, say, car accidents, much less car pollution in cities) as compared to being saved from death.
No, I am serious. And it will get much worse in the future. Check out [1], [2], etc.. And lets not forget the increased storms, flooding, etc. We are all responsible for that. But those that use electricity like big tech should be held especially responsible because they encourage the bad behaviour and use resources directly.
> Even outside of the scientific fields, AI sure is solving a lot of my problems, saying one is an "enemy of humanity" is the sort of hyperbole I'd only see on HN.
Solving your problems doesn't really mean solving the world's problems, just like making the top 10% richer doesn't make the world a better place. I absolutely consider them the enemy of humanity.
1. https://ourworldindata.org/part-two-how-many-people-die-from...
2. https://www.weforum.org/press/2024/01/wef24-climate-crisis-h...
>I keep getting told by friends not to use AI because it is destroying the environment.
I side with them. If you're environmentally conscious you absolutely cant support AI. The amount of energy and the backing of basically every AI is fossil fuels.
You should oppose AI use.
Let's not also forget material use. According to [1], "Global AI demand is expected to consume 4.2-6.6 billion cubic meters of water by 2027, surpassing Denmark’s total annual water withdrawal of 4-6 billion cubic meters." And there's also all the mining required for the materials to build the computers, and the fossil fuels shipping them and making them in the first place in places like Taiwan.
Reference:
1. https://unric.org/en/artificial-intelligence-how-much-energy...