Meanwhile, AI is getting more efficient and will no doubt continue to get more efficient. Research that undermines the need for huge datacenters can’t come soon enough.
Higher efficiency usually just means more use/exploitation, especially in capitalist countries
The capitalists are pushing generative AI, regardless of whether it’s good for the environment or actually useful. Ideally, if it’s a net loss for humanity, it would be halted, at least until that can change. But since there’s a lot of capitalism doing its thing on this with nobody putting a stop to it, in the short-term, the best case we can hope for is that increasing efficiency at least makes it environmentally-friendly and dilutes the reliance on major corporations using huge datacenters. If it got to a point, for example, that you could do your own model training experiments and run large models on a smartphone, that would at least put generative AI more in the realm of coding where knowledge is the main barrier rather than compute and it would be harder for capitalists to control what is happening with it and how it develops (which we know is not going to be for the good of the working class).
There’s also the fact that China is a significant force in the development of AI, both generative AI and other kinds, and I would hope they’re being more conscientious about how they’re developing it and integrating it than, say, the US. But the point being that with them also working on it, it’s not like we could just stop the capitalists and put the AI back in the bottle; the socialists are working on it too, in their own way.
What I meant to say was that these data/computing centers exist, so they will be used. If any efficiency gains are integrated into western LLMs, there will be more output from a given data center.
It’s like what we see with gains in energy efficiency; instead of energy use being reduced, the thing is typically used more.
Depends on what kind of thresholds get passed I think. Yeah, they’re not gonna immediately start dismantling datacenters if LLMs get efficient cause there’s money invested in it, but well, consider this part of the article:
The day before the Stargate announcement, Trump’s inauguration date, the Chinese company DeepSeek launched its own AI model, claiming it had used far less computing power – and therefore less water – than its western rivals.
More recently, Bloomberg has reported that Microsoft is pulling back on some of its plans for new datacentres around the world.
Are those two events linked, Deepseek’s release and Microsoft pulling back? I don’t know for sure, but it’s possible. Why invest as much into it if the projections start looking like they won’t be needed. That’s where I’m coming from on this.
So far the big thing with generative AI has been that they take ridiculous amounts of compute (GPUs) to train and do inference (generate) on a trained model, with the prevailing belief being that the primary way to keep moving the needle in model quality is to keep throwing more compute at the problem. Deepseek put that more into question, doing more with less (relative to the best out there).
But I don’t want to get more efficient I want Project Stargate
I assume someone will deal with this water transferring apparatus appropriately at some point, tbqh…
