[____mr____]

If the tech is here to stay, my question is: how and why? The how: The projects for the new data centers and servers housing this tech are incredibly expensive to build and maintain. These also jack up the price of electricity in the neighborhoods and afaik the US electrical grid is extremely fragile and is already being pushed to its limit with the existing compute being used on AI. All of this for AI companies to not make a profit. The only case you could make would be to nationalize the companies and have them subsidized by taxes.

But why?: This would require you to make a case that AI tools are useful enough to be sustained despite their massive costs and hard to quantify contribution to productivity. Is this really the case? I haven't really seen a productivity increase worth justifying the cost, and as soon as Anthropic tried to even remotely make a profit (or break even) power users instantly realized that the productivity is not really worth paying the actual compute required to do their tasks

[ttoinou]

Do you need a measure or a quantification to do anything in life ? I don't wait for others benchmarks or others computing ROI factors to actually start using a technology and see it improves my workflow

  how and why?

How : we'll always be able to run smaller models on consumer grade computers

Why : most of the tasks humans need to do that computers couldn't do before, now can be improved with new AI. I fail to see how you can not see applications of this