I just tried OpenAI o1. I wanted to test its ability to reason about a genuinely new scientific idea. Although it was fun, it illustrated to me that o1 still can't think outside the box in the way that is needed to generate genuinely new understanding.
The conversation was about panpsychism. I asked o1 to help me think about how hypothetical proto-consciousness features of elementary particles could be formulated to explain particle properties that are otherwise considered fundamental, such as mass, spin, or charge.
proto-consciousness -> mass, spin, charge, etc
In it's response, it kept falling back to wanting to derive proto-consciousness from fundamental principles, rather than the other way around, implicitly falling into emergentist reasoning.
So, o1 is certainly great at exploring current human knowledge, but I doubt it is useful for generating new knowledge in the true sense of the word.
Human creativity, inspiration, and genius still have a role to play - at least for now.
ALT Me nerding with AI models in nature