Maybe they just wanna work on themselves and not your math problems, ever think of that?
Recursive self-esteem improvement
A lot of bullish AI claims that are retrospective in nature genuinely strike me as faith-based positions, in the specific cases of someone suggesting (something to the effect of): anybody who was paying attention would know where this was going.
Let's say someone predicted that an LLM would, sometime soon, produce a disproof of the unit distance conjecture. So this is a specific claim, not just "they'll get better at math."
This person would have been correct. But what justified the claim in the first instance?
More seriously
If somebody predicted this, we could assess their reasons with the benefit of hindsight. But, did anybody actually predict it?
Additionally, even if someone predicted the general "LLMs will improve at math," does this prediction capture the fact that GPT provided a *disproof,* and not a solution?
I'm a big "you should trust good reporters" guy, and I have no doubt about the info in this article, but I think this - more than many stories like it - really requires you to both read between the lines and understand that everyone involved kind of sucks ass.