This is a great talk by Chollet about the path to AGI and why scaling alone (and not even scaling + CoT reasoning) will get us there. Video link below. He discusses abstraction, symbolic reasoning, continuous vs discrete optimization (program search), and many other important topics. Note he keeps using the term "Fluid Intelligence". $trillions and geostrategic dominance are on the line so AI researchers have to be realistic about what "intelligence" is. They end up recreating the methods of psychometrics (now applied to AIs instead of humans) that have largely been driven out of the academy by woke stupidity.

Humans do discrete combinatorial thinking that current LLMs are not good at. In particular, humans can combine discrete modules with abstracted properties to build mental images of complex programs. Ofc the human brain may be using some kind of continuous "distance metric" heuristic in the background to make guesses / have intuition about this discrete combinatorial process. Chollet suggests that we need to build the capability for the AI to embed the combinatorial problem (with abstracted properties for the modules) into a latent space with a metric so as to use continuous optimization methods to find solutions (ie exploiting intuition and avoiding brute force search).