It's just really weird to me that people are all shocked-Pikachu about Carmack wanting to be the auteur in this scenario and to wield Meta's effectively-infinite resources in the exact ways that he wanted. It's apparent from every interview and article about him that this is how he operates. It seems to have worked well for him in the past, but it was obviously going to be a major culture clash at Meta unless they gave in and let him run the thing.
He was likely to have more success at Meta than his AGI startup. When it comes to intelligence and cognition no one even knows where to begin. The same for studying the brain. And the neural networks in the brain don't resemble anything like the neural networks in current AI.
I guess there's this idea that we'll wander into the right territory. That might work for other things but probably not for the most complicated organism on our planet.
Are those papers breakthroughs in understanding human cognition? It feels like there must be some philosophical underpinning to creating human-like intelligence.
I suppose there are two approaches: 1) understand the brain in all its complexity 2) wander upon something that seems like human cognition but isn’t (i.e. GPT)
Carmack and everyone else is taking the latter approach. Carmack may end up building something that seems intelligent — if that’s what you mean by intelligence.
Consider Chomsky’s view on current AI. He may disagree with me but he certainly disagrees with the idea that actual intelligence or something like AGI will result from current efforts.
> It feels like there must be some philosophical underpinning to creating human-like intelligence
Cognitive science, mostly stemming from this common intuition, has failed us after spending decades of research effort, while minting more than a few academic careers.
Same with many once prominent public intellectuals.
GPT is certainly presenting itself to be very uncharismatic to most and humiliating to some.
