Maybe nothing. Maybe everything. Assuming truth (a bit early in the game for that, but this is a discussion, not a Board meeting to approve $500 million in expenditures), it could upend spending plans for many of the world’s largest tech firms. It could change the direction of data center construction. Like so many things that are unknown into the future, it could change how hundreds of thousands of companies interact with customers. I really don’t know, maybe it’ll put an end to the Comcast call centers in Mumbai if AI can inexpensively deal with US customers here.
I’d say “I really don’t know” again, but I hope that’s clear. Bigly promoted things often lead to nothing (Hello Google+ and Threads) and little things often have life changing results (the Mouse, Post It notes, Netflix). There’s an interesting post on Saul’s about Jevon’s Paradox, the phenomenon by which something becomes more efficient but doesn’t decrease the demand, it makes it more usable in more contexts and increases ubiquity. The beginning was Jevon’s observation that the steam engine used coal more efficiently, but rather than coal use going down it went up because steam power was started being used so widely after it became economically viable to do so.
So what will “cheap and ubiquitous” AI presage? No idea. A big hurt for Nvidia and others who have been leading to this point, but that’s micro. Macro? Give it 10 years, we’ll all find out.