Reposting this answer to a question from @geoffreyhinton about whether I think LLMs "understand" what they say.
I point out what I think is missing from current architectures to reach cat-level intelligence (never mind human level): world models and planning/reasoning abilities.
AGI
-
LLMs Understanding: Missing World Models and Planning Abilities
By
–
-
Rethinking General Intelligence: Beyond Human-Centric AI Assumptions
By
–
That's why we don't think that human like intelligence is the on,y form, and also why I think "general intelligence " is a stupid name based on Tue wrong assumption that human intelligence is general.
-
A* vs MCTS: Planning Algorithms Explained by FAIR Researcher
By
–
Yuandong is one of several folks who have been working on planning at FAIR.
He explains the difference in applicability between A* (search for shortest path in a graph) and MCTS (search in an exponentially growing tree). -
Yuandong’s Planning Research Approaches at FAIR
By
–
Exactly.
Yuandong has been working on various approaches to planning at FAIR. -
Building and Training World Models for AI Systems
By
–
Surely. The question is how to build and train this world model.
-
Brian Greene AGI Debate AI Safety Interview World Science Festival
By
–
of my public interview with Brian Greene at the World Science Festival in NYC a few weeks back.
It is followed by a debate about "AGI" with @SebastienBubeck It ends with a debate about AI safety with Sébastien and Tristan Harris.
I find it strange that Tristan uses the -
Questioning AI Progress Claims: Closer Than Believed?
By
–
How could I convince you that the answer to (a) is "near, perhaps, but not as near as you might have been led to believe", and that the answer to (b) is "no"?
-
AGI Timeline Predictions: 2015 Optimism Proven Wrong
By
–
The self-delusion was to claim in 2015 that AGI was just around the corner and that a non-profit organization was going to reach it alone before anyone else.
That all turned out to be wrong.
And we're still some ways away from human-level AI.
Generations after generations of AI