knotthebest

Do note that Terry has access to the full o1. o1-preview is, well, a preview.

They are, though. I spoke to an author just yesterday. They did mostly use brute-force.

Still, very marginal improvements. There is also Amdahl’s Law which limits the speed gains of parallelisation.

Skill issue tbh

You’re not understanding the tests. They were designed for humans. They were also designed to be taken once. LLMs have been trained on these tests numerous times. LLMs are also not humans. We can’t conceivably compare the two

Well, they were trained on those tests. What LLMs literally do is map inputs to outputs. That’s why they can’t do much outside of training data.

I just don’t think Ray Kurzweil understands these systems well enough. His views aren’t of much substance anymore.

Nothing particularly interesting to my knowledge. You should read this though, some people at Symbolica AI worked on it too: https://www.google.com/url?q=https://arxiv.org/abs/2402.1533...

I’m not sure how this would solve the hard problem. Could you please elaborate on that?

My guess too

I think I may be missing something. Why would you be interested in the automation of theorem production? Wouldn’t this make mathematicians obsolete? How far away do you think we are from that?

I ask as a newbie in math; math is a passion of mine. I am genuinely reconsidering going into math research as I fear just being automated away.

At which point would mathematicians become obsolete? Something like this seems like it could automate a lot of mathematics research, no?

I’m not so sure of this. The fear generally isn’t that the model itself is dangerous, more so that how it is used by someone might be.

Eh

Generally just curious about what others a bit more knowledgeable in this field than me might think.