I quite like Codeberg’s stance on LLMs and think they make some valid points. They also raise real concerns that we should probably take into account in our personal/national approach to LMMs.
It is concerning that there is so little call for restrictions and proper safeguarding on LLMs and as i’ve mentioned before it’s massive negative impact on environment and cost of living.
Environment - The media writes about how how much water is “wasted” when cooling data centers. No water is wasted - it evaporates and then eventually comes back as rain.
Now it is true that there can be a water relocation problem with such data centers. Therefore it may not be advisable to build such a center in locations which have a marginal water supply to start. But there are plenty of cities which have an abundance of water; there is no environmental issue using that water to cool data centers. To the extent that such a data center may use drinkable water, the politicians should be certain that for perpetuity the data centers have to pay their accurate fair share of the cost to replace that drinking water.
Cost of Living - There is a lot of political discussion currently about data centers raising the cost of electricity and thus raising the overall cost of living. The answer there is clear - local politicians (of whatever stripe) need to grow a spine and be sure that when a data center is built, the company operating the data center needs to pay their accurate share of the cost of generating electricity - and that needs to remain as such in perpetuity. Problem solved.
On both issues you would be correct to say that my analysis is correct from an engineering perspective but challenging to achieve on a political level. That woudl be fair. But call a spade a spade - the enemies are not AI or data centers. Rather the enemies are politiicans of any/all stripe who turn a known engineering solution into a political football…
I’m not sure I would fully agree with you although politicians being culpable and giving preferential treatment to AI corporations is an issue.
The broader aspect is that AI is using limited resources at an exponential level and extracting these resources has a negative effect on our environment. I would also suggest the AI is responsible for the massive increase of technology costs as a whole due to its unsustainable consumption of RAM, semiconductors and other hardware, which massively impacts cost of living as technology is embedded in nearly every aspect of society. Apple’s increase of hardware by 20% is a case in point.
It seems to me that the full impact and ramifications of AI have not been properly thought through by governments and its environmental impact completely avoided.
In terms of water there is an argument that the industrial level water required by AI is denuding water tables that have taken centuries to fill. This is not easy to replace and of course the rain is unlikely to fall on the area that uses the water.
In terms of electricity the high usage required by AI data centres overloads local systems meaning they need upgrading and passing that cost onto locals is unfair as their usage was not overloading the system.
I’m amazed at how toothless the push back is by the green party. It just demonstrates that political status is more important than damage to environment. What bugs me about all this is the penny pinching carbon zero we’ve had to put up with for a decade+ that pales into insignificance of AI’s carbon footprint and not one political party makes a realistic comment.
I know some people doing the AI rollout for Friends of the Earth (UK). Their conundrum is simple. AI is energy intensive, yes, however it also allows things to be created or done that were simply impossible before (at a given investment point). They can monitor, critique or innovate in ways they couldn’t before. So it’s a balance between using (and promoting) a technology that could do environmental harm, vs the fact that it furthers their social cause and therefore contributes to a better planet.
Essentially the difference is between gross impact and net impact of AI. Yes AI uses energy and resources, but the real question is whether it’s a socially positive or negative use, considering everything in the round. That’s a complex question to answer, and the Greens recognise this fact too. IMO their focus should be on greater regulation of tech companies, not AI specifically.
Also, the issue with RAM is simply that it’s been dirt cheap for years now and so there was no profit in making more factories to produce it, as supply easily met demand. Hence we had a fixed supply. The RAM shortage is simply that demand spiked quicker than supply could meet, since new factories take a couple of years to build. Given that it’s now profitable again, more factories are being spun up to meet the demand. Hence in about 3 or 4 years, we’ll be swimming in RAM again. So it’s not really a fundamental AI issue, just a short term supply one.
Limited resource - Compare the harm that burning fossil fuels does with the “harm” of using water for cooling a data center. It is a non-issue as long as the data center is placed somewhere that has an abundance of water. There are lots of places in the USA and elsewhere which have a sufficient water table that the use of water for this purpose is a negligible issue.
Electronics Shortage - This is a short-term issue. The free market will do a fine job correcting this.
It seems to me that the full impact and ramifications of AI have not been properly thought through by governments and its environmental impact completely avoided.
This has never happened in world history so there is no reason to believe it would happen now. We cannot even get world governments overall to cooperate ideally for public health purposes - the ultimate example where cooperating is in both in the world’s interest and in self-interst.
Nice idea, but the reality is that is less than 10% of AI usage. The rest is used for commercial advertising algorithms, algorithmic trading, consumer chatbots, and speculative search features.
This is a moot point in my opinion. There are not enough resources going around. Estimation is that AI use of copper is affecting renewal energy’s need for copper. It doesn’t matter if it’s for positive use if it’s not sustainable for our existence. This includes your comment on RAM. There is not enough raw materials to go around.
The point is that this is not the criteria. In reality, they are built where tax breaks, cheap land, and fiber connectivity intersect. In the US Arizona, Utah and Virginia are extremely popular and they certainly don’t have an abundance of water.
Not at the moment. Sustainable energy cannot keep up with need. Apart from nuclear energy there seems no viable solution.
I’m sorry, but I don’t even know how to engage with this speculation. There are so many ‘firsts’ in history we should probably learn from it. The Montreal Protocol banning CFCs is probably a prime example.
That is an irrelevant comparison though: we should be transitioning away from fossil fuels, even if we build no more data centres.
And the water is only available for evaporation once a process of decontamination has been used: are you confident that the companies behind every data centre does or will employ those measures?
The World Economic Forum Report on Global Risks identifies ‘adverse impacts of AI technologies’ and ‘Biodiversity loss and ecosystem collapse’ as significant risks in the next 10 years.
The water demand of AI technologies is likely to threaten global and national water security, especially in areas of existing water stress, which can in turn threaten the biodiversity of local areas and the needs of human populations.
Demand for fresh water is expected to exceed supply by 40% by the end of the decade and 55% of global data centres are in river basins with high risk of water pollution, meaning much of the local water may be unsafe for use, increasing the pressure on clean water supplies and worsening the overall water scarcity in the regions.
Nearly 68% of data centres are near protected areas or Key Biodiversity Areas which rely on clean water supplies for the health of the ecosystems and communities that depend on them. Without it, these areas face greater risks of habitat loss, species decline, and reduced capacity to support both nature and people.
Now, it does suggest some mitigating actions, though a) these mostly apply to new developments, not existing ones, and b) require concerted government action by government both competent and caring enough to act.
I suspect neither you nor I think that the current US government – or many others, for that matter – is likely to become a sudden convert to environmental protection. Does that mean we should all shrug our shoulders and stop pointing out who is responsible for the risks? How else can we influence the politicians to make sensible choices, other than by vociferously challenging the AI companies, and the politicians who enable their destructive tendencies?
Let’s please cool down the data center discussion a tad.
This was my favorite part of the Codeberg writing:
It seems like many ‘vibe coders’ don’t realize that they don’t actually have a community around them. They build projects as if they had, and spend resources accordingly. We see projects having a lot of code activity, heavy CI/CD testing, frequent and large release binaries. Sometimes, it feels like the amount of supported platforms exceeds the amount of actual users.
To us, it seems ridiculous to see projects with a single developer and virtually no users consuming as much or even more resources than some of the largest community projects on Codeberg, which operate frugal with CI/CD and storage resources. We do not believe it is reasonable for Codeberg to invest our precious donation money into hosting of large ghost projects.
I’m seeing this everywhere, and been guilty of it myself. It’s too easy to keep hitting the code button before you share with people now.
Imagine an open source host that would only let you host a short essay and a thousand lines of code until five people replied that they’d like to try it. Codeberg and Github aren’t going to do that, but that was closer to the previous social contract.
@svsmailus Those are some very insightful observations.
Unfortunately the likelihood that world governments will approach anywhere near that level of factual reasoning is nil. We cannot even get a global consensus on the benefit of vaccines to everyone.
My 10% figure is an informed assumption based on who actually owns and operates global AI infrastructure. Large commercial cloud providers dominate global AI compute, while NGOs are estimated to use well under 1%. Even when you include universities, humanitarian agencies, and other mission‑driven groups, their combined footprint is still small. So it’s reasonable to conclude that less than 10% of total AI use is directed toward the betterment of humanity and the environment rather than profit.
Part of this is probably the fact that AI is rather congratulatory about even the most pedestrian of ideas and accomplishments. I’ve seen at least one person get talked into the viability of a business idea by AI. Not from AI providing data, but from AI slathering them with superlatives until they believed their idea was going to be The Next Big Thing™.
We need an AI that looks at peoples’ vibe-coded apps and says, “um…hey Dave…is there REALLY a market for this?”
80% of California’s developed water supply is consumed by agriculture in its water-scarce regions, such as the Central Valley.
Two of North Carolina’s major watersheds—the Cape Fear and Neuse River basins—have been materially degraded by the runoff from the high concentration of CAFOs (Concentrated Animal Feeding Operations) in the region—hogs especially.
We don’t need to grow crops and raise livestock this way; we’ve chosen to. (CAFOs are industrialized animal cruelty. They’d be morally repugnant even if they had no deleterious environmental impact.) Infrastructure, subsidies, water law, insurance, processing capacity, and livestock markets have made environmentally costly production choices economically rational for individual producers.
I’m not unconcerned about the environmental impact of AI; I’m frankly more concerned about the baked-in lunacy of US industrial agriculture. I’ve been heartened by the US backlash against data centers at the local level (despite the fact that that opposition isn’t always grounded in fact) because it offers at least a glimmer of hope that this time we won’t let the degradation of our commons be the path of least resistance.
I was talking about the reason that FOTE use it, ie their own use causes environmental harm and their own use also offers a social purpose, so they’re balancing the two. They’re not so bold as to assume they’re good and everyone else is bad. The point is that this balancing act and judgement call might be why the Greens also aren’t “anti” AI, they also benefit from its use.
That said, the idea that all people who are not universities, humanitarian agencies etc are using AI for entirely negative purposes is a gross oversimplification. it doesn’t take much imagination to come up with positive uses outside of those purposes, at every scale. The truth is the net impact is essentially unknowable, not least because AI is in is infancy and the biggest impact is yet to come. I think its presumptuous to call the entire thing a bust already.
I would agree with you. But it is a gamble and a rather large one at that. It does feel like all eggs are going in the AI basket. Let’s hope it doesn’t fail.
How can it “fail”? There are already many clearly useful applications of AI which have transformed industries. Yes the pricing and profitability parameters need to be tweaked - but the free market will take care of that.
Has it though? I don’t think we have enough data to support this yet.
This assumption is overly optimistic. I have far more faith in the ability of big tech corporations to manipulate the free market for their own profit than I do with the ability of the people to use their wallets to drive the free market towards outcomes favorable to themselves.