Not too many people have a problem with AI technologies conceptually, and arguing like they do is ignoring the real criticism in favor of semantics. People have a problem with the economics of how AI things are being implemented, positioned, marketed, and used. Burning data centers would radically change the economics of AI.
reply> People have a problem with the economics of how AI things are being implemented, positioned, marketed, and used.
replyThose economics are also changing very quickly, with free local AI becoming increasingly dominant for many everyday uses and even starting to become relevant for the enterprise ones.
How many devs would be able to keep working if GitHub disappeared tomorrow?
You can do inference on SOME laptops, but the current shape of GenAI need massive data centers to be used at scale.
replyAlso the existence of various big tech companies rely on these data centers being place, without them they are useless.
The nice thing about local AI is that it really can run anywhere, you just need enough storage space for the weights and the context. It just gets slower if you run it on potato-level hardware.
replyFor bang-on user cases like coding, sure. For concept art and other still-image diffusion tasks, sure. For damn near anything else, calling hardware that doesn’t approach the inference breadth and performance of data-center-hosted remote services ‘potato-level’ is pretty disingenuous… never mind any significant training. Not only that, hardware is less accessible than it’s been in years— nvidia is re-releasing the 3060 so gamers can buy something. For anything you’d stake your business on? Good luck.
reply