alt Hacker NewsI would disagree regarding the opening paragraphs about intelligence, but then we don’t have a generally agreed definition of intelligence. There is a scientific view that a lot of what the human brain does and has biologically evolved to is predicting what will happen next, and next-token-prediction is reasonably related to that.
However, what seems to still be lacking in LLMs — aside from plasticity (continuous learning), perception of time, and anything related to emotions and desires (which underly motivations) — is various types of awareness, and sound judgement. We are still far from a good understanding of these aspects.
LLMs are like a static snapshot of a brain. Inactive till you run one tick with it. I expect AI models becoming much more dynamic. The model needs to be live updating itself. Then it will become a living thing. But hardware wont allow for that yet. Maybe first small dynamic LLMs are possible.