I've built a mathematical universe, a growing library with over 20k definitions, theorems, and examples, and over 100k dependency edges.
https://t.co/0CeH9gansq
GPT-6.2 Astra—a frontier model built for your hardest projects. It sets new highs on evals across professional work and scientific research at less than a third of the cost per task of GPT‑6 Astra.
You’ll see the difference in things like slideshows, websites, and 3D designs.
@thomasfbloom Personally I would credit AI and describe the capacity in which it was helpful, but I also understand why some people might choose not to. It really comes down to treating AI usage as a spectrum, there isn't a universal standard to what amount of AI usage requires disclosure.
@thomasfbloom I'd also like to point out that AI-assisted research does not imply zero human effort. It really depends on how AI was used. You could guide AI to explore in a direction of your choice, by asking questions that kill bad ideas and potentially save time and effort.
@HarryStoltz1 As great as formalization is, it’s not without points of weakness. Someone (human or AI) has to translate mathematical statements into LEAN, and there’s always a possibility of mistranslation.
@noahlt Because advanced theorems and definitions often require long chain of prerequisites to even understand their meanings. That’s why I built a math library with fully explicit dependency relationships