There's an excess of GPU power and data centers right now and a lot of companies are buying that up to train physical robots to do physical jobs. This is the next wave of innovation.
Yeah, sure, but there have been attempts to build self-driving cars for much longer than the current wave of LLMs capable of doing knowledge work, and we're still not quite there yet. If you think of it, it's pretty obvious that it's easier for AI to manipulate bits and bytes rather than interacting with the physical world. So, while actually intelligent robots might be a good thing (doing jobs that you can't find enough people for, or jobs that currently are not realistic, like going through all the trash a city produces and sorting out the recyclable stuff), I'm not holding my breath waiting for them...
100% agree but there's a big difference happening today. Investors are seeing the huge payoff for NVDA, OpenAI, and Claude and want to pour their money in the next compute heavy idea. That just happens to be physical automation right now. Money plus competition usually equals leaps forward in innovation.
Self-driving cars and LLMs have a lot in common these days. The same wave is driving both forwards. I don't quite know what you mean there are not quite there yet. They are literally doing just that: driving customers autonomously around in large busy cities. Is perfection the bar to clear?
Yeah, I know, autonomous cars are held to higher standards than human drivers, but still, I think a really autonomous car should not depend on Google painstakingly mapping out every street before being able to roll out Waymo in a certain city. It should be able to drive safely in unfamiliar environments too, same as humans can.