Here is my personal statement on AI. There is no question that current frontier LLM models are powerful enough to generate proofs of research level problems. An impressive record of AI-claimed results is here.
But mathematics is not just about formal proofs. To quote Bill Thurston:
« The product of mathematics is clarity and understanding. Not theorems, by themselves.”
To quote Alexander Grothendieck:
« La création se distingue de la production par le fait qu’en plus de l’oeuvre exterieure elle s’accompagne d’une oeuvre intérieure qui en constitue l’aspect essentiel ».
In other words, through LLM generated maths, you’re missing the whole point of the creative process which is about the internal journey, the slow growth of new concepts, and, ultimately, the joy of communicating new results and ideas to your peers.
Due to the current cynical behavior of US based AI companies, i will indefinitely boycott them. Their clear purpose is to destroy the academic system and ruin human knowledge.
At the moment, i will stop posting papers on Arxiv due to the fact that they are used without my consent to train LLM’s. The famous preprint depot Arxiv, once the pride of a community, is now rotten by a flood of AI-generated Junk. My research will only be communicated through talks and printed papers will be sent privately to trustable researchers only.
I have joined the Association for Human Mathematics whose objective is to protect mathematical knowledge against the threat of soulless AI companies. In particular people involved are thinking about secured, alternative ways to share research without feeding AI bots and Theorem hungry Users/Prompters.
I still have faith in the possibility of a well regulated, open source AI model would could benefit scientific research and mankind in general. I just don’t see it coming from those profit oriented companies.
