If OpenAI resolved the Riemann hypothesis by finding a nontrivial zero at 0.50000003 + 3531696584231.17843174i, that would be very cool and probably very impactful. But it might not necessarily demonstrate model capabilities beyond those which have already been demonstrated, especially if the session that found it was one of a thousand launched to explore different areas of the problem space.

More generally, do you expect that there's some capability threshold where people will no longer study or analyze AI model outputs, and instead just sit there slack jawed saying "so cool!" every time OpenAI announces novel ones? I don't really understand why that would be or why someone would want that. If you're interested in the pure experience of a complex machine outputting satisfying results, I'd recommend getting into sports cars.

It would be my reaction if we're discussing a blog post from OpenAI, yes. I would be looking at it extremely critically, wondering what they're misrepresenting to make it look cheaper, easier, and why they're trying to make it look like only their model could possibly do this.

Look at their recent claims about their model "escaping" - there was literally a Guardian article calling them out for being hyperbolic! Again, it wasn't that they lied, their marketing department is too savvy for that. They just present it in way that's, well, marketing.

As for the actual result, I'll look for secondary posts by actual mathematicians and draw my conclusions there, not from this marketing blog post about results from a secret model.