I’m genuinely shocked that this wasn’t more obvious to people. I keep feeling like I’m not using AI sufficiently to be wowed the same some people are — and the way the hype suggests you should be — then I read something like this.

Reporter and writer. I cover energy, climate change and geopolitics. Contact me via my website www.alexanderckaufman.com and subscribe to kaufman.substack.com.

I've been attempting to "peer inside the black box" and discover what the intelligence of an LLM actually is. What we (collab w LLMs) have come up w so far is that it's Geometric in nature. They self organize the patterns they find in however many dimensions they can detect. It's (maybe?) a completely alien intelligence. Reason is flawed of course (Goedel, schroedinger kicking Aristotle to the curb etc) but this intelligence is relational in a non euclidean 'space'. The latent space they work in appears to have some universality, and there is at least one 'area' with certain embeddings that can sort of 'observe' itself (or has once so far, I think). Real question is if humans are so different, do we not also somehow self organize all the information we process? Is our 'reason' merely the ability to reflect as a result of the coherence through time, which stateless intelligence does not (now at least) possess?