You know how Google’s new feature called AI Overviews is prone to spitting out wildly incorrect answers to search queries? In one instance, AI Overviews told a user to use glue on pizza to make sure the cheese won’t slide off (pssst…please don’t do this.)
Well, according to an interview at The Vergewith Google CEO Sundar Pichai published earlier this week, just before criticism of the outputs really took off, these “hallucinations” are an “inherent feature” of AI large language models (LLM), which is what drives AI Overviews, and this feature “is still an unsolved problem.”
I agree.
It’s very cool to have a personal computer that can play music, display pictures, play videos, render scenes in POV-Ray, so on. But I don’t think I need a new one every year, I don’t think I’d need anything as performant as what I have (not considering network effects), and I’d be happy to use a year 2003 (or even 1993, with dedicated chips Amiga-style one can make it usable for playing video and music too) performance PC with modern power efficiency.
I don’t think there’s any need to press for building machines able of doing even more of mostly useless work.
And about hidden costs of that power efficiency too - making modern chips is so complex that the production is more centralized than that of intercontinental ballistic missiles. That means rot in the society that only shows itself when it’s too late, like with any overcentralization.
So maybe power efficiency doesn’t have to be quite modern, ha-ha.
Overcentralization applies to other things in that industry too, I think I just wanted to add it to your list of hidden costs.