Back

I saw M3GAN, it’s a fun and witty techno-horror (️️️️). But it’s got some pretty pointed mockery of the #LLM crowd. Don’t get me wrong, I’m 98% sure you can’t turn an LLM into a general #AI just by throwing more compute at it, no matter how many feedback loops you throw in there. You need multi-modal symbolic reasoning. They’re snake-oil salesmen. But if they weren’t, that’d be worse. Because they’d be phenomenally irresponsible: “You installed a learning model you could barely comprehend…”