>>/1578/
> normie local shit
Time to DM those normies and get into normie things :)
Actually I know what kind of content you're probably talking about and it's always SO AWFUL, like mind-numbingly bad
I've been reading a pretty interesting podcast transcription about an AI prize for a certain dataset (ARC) and I think it's a very good read.
Made me seriously reconsider if there's any chance of getting AGI with current approach. Well, I was always skeptical but it went from maybe if they scale it big enough and throw enough resources at it it might work but it's a very small chance to the chance is so remote that it's unfeasible most probably.
Also it really illustrated well what modern AI can and cannot do, it's very good at memorizing patterns and spitting them back at you and abysmal at trying to reason through something that's not in its dataset, even if it's literally a 5-year old problem.
It did solidify my opinion that it's a tool at best and only a good tool if somebody already solved a very similar problem to the one you're providing it on the web (so basically a very sophisticated search engine)