Google DeepMind CEO Says Some Form of AGI Possible in a Few Years

Re:Google DeepMind CEO Says Some Form of AGI Possible in a F... by nextos on Tuesday May 02, 2023 @10:26PM (#35796990)

There's a good summary in the Introduction to Kevin Murphy's second tome [1]. He quotes Josh Tenenbaum, who said: "Intelligence is not just about pattern recognition and function approximation. It’s about modeling the world".

I agree, and I think Goodman & Tenennaum [2] is a great place to see other things that may pop up in the road to AGI. LLMs are great, but they do too much at once. I think moving towards AGI requires some form of symbolic reasoning, possibly combined with LLMs, which may play the role of intuition and kitchen-sink memory.

[1] https://probml.github.io/pml-book/book2.html

[2] https://probmods.org

[ Parent | on HN ]

Re:Google DeepMind CEO Says Some Form of AGI Possible in a F... by willbudd on Tuesday May 02, 2023 @10:42PM (#35797117)

Modelling the world is function approximation: the world is the function and the model is the approximation.

[ Parent | on HN ]

Re:Google DeepMind CEO Says Some Form of AGI Possible in a F... by NumberWangMan on Tuesday May 02, 2023 @11:59PM (#35797780)

Do we know for sure that LLMs can't possibly be doing symbolic reasoning of some sort, internally? I thought we couldn't really interpret them yet.

[ Parent | on HN ]

Re:Google DeepMind CEO Says Some Form of AGI Possible in a F... by stevenhuang on Wednesday May 03, 2023 @06:35AM (#35800522)

There's mounting evidence LLMs do appear to form models of the real world https://thegradient.pub/othello/

[ Parent | on HN ]