John Carmack's 'Different Path' to Artificial General Intelligence (2023) | Hacker News Reader

John Carmack's 'Different Path' to Artificial General Intelligence (2023)

dallasinnovates.com

4 pointstosh3 comments

>We’re in the midst of a scientific revolution right now, because 10 years ago, there was not the sense that AI was working. We’ve had these AI ‘winters’—a couple of them over the decades, in fact. It’s funny, because virtual reality also went through this: It had almost been a bad word because it had crashed so bad in the 1990s, people didn’t even want to talk about it. And artificial intelligence had a couple of those cycles where hype spins up, money flows in, it underperforms, and then it crashes, and nobody wants to talk about it. But this last decade was different, and people who don’t notice how different it is this time aren’t paying attention, because the number of absolutely astounding things that have happened in the last decade with machine learning is really profound.

Is it that different?

Except that the overal IT and comsumer computing industry is 1000x bigger and thus the money involved and the available hype too?

Without some major breakthrough in LLMs, the current hotness will also fade, and the money will move elsewhere. We have a couple of years for that, judging from similar cycles (first they get to bundle AI to everything, and sell LLM based AI to the enterprise and consumer -- then the market dies of, and people start thinking "that was all there is?", that being more or less what current LLMs do).

"And it feels like we are a half-dozen more insights away from having the equivalent of our biological agents."

And

"If you took the things that people talk about—GPT-3, Imagen, AlphaFold—the source code for all these in their frameworks is not big. It’s thousands of lines of code, not even tens of thousands. Now, they’re built on top of a big framework of supporting ecosystem stuff, but the core logic is not a big program."

I'm with mr. Carmack there. He regards today's hotness like LLMs not as pointing the way, but as pieces of the puzzle. We need a few more pieces. And the overall picture probably won't be too complex. Big (as in: scale), probably. But not very complex.

We just don't know (yet) what pieces are missing, how they fit together, and how to get there. But (at least to mr. Carmack) it feels like we're within arms reach of that.

I think he's right. And maybe he'll come up with one of those missing pieces? At least to some people that's worth a $20M gamble.

"And he gave me a list of like 40 research papers and said, ‘If you really learn all of these, you’ll know 90% of what matters today.’"

That alleged reading list given to Carmack by Sutskever is here: https://news.ycombinator.com/item?id=40297984

Reply on news.ycombinator.com