Very few people, in history, have correctly anticipated what form of new technology would emerge and which transformations it would entail. We don’t know (nor are we desperate to guess) what long-term changes AI will produce in mathematics. Precisely for this reason, we think it's healthy that some will keep doing mathematics without outsourcing any form of thinking.
We believe, indeed, that some risks connected with the diffusion of AI in the mathematical life are quite concrete. Among them, multiplication of technically dubious articles, dilution of good content, flattening of taste and style, but also relaxation of social values that are peculiar to the community: care, patience, rigor, honesty. Yet even if AI were to become infallible, we would still find it safer that someone keeps an exclusive contact with an unmediated mathematical practice. We believe, moreover, that it would be sound, for the sake of cultural continuity, that some younger mathematicians preserve such a contact.
This is not, for us, the same kind of delegation involved in using a numerical software or a search engine, tools automating tasks that a (not necessarily brilliant) human could, at least theoretically, perform with enough time. AI risks ending up substituting (or faking) the very act of thinking as well as eroding the social habits that sustain and flesh out mathematical life. And it is that act and those habits that we choose to keep intact.
We recognize that the boundary between what is usually meant today by “AI” and more conventional tools (like Google Scholar or a spellchecker) is not always sharp, and we don't aim to produce a tedious taxonomy. Rather, without in any way considering it inappropriate for others to use the fanciest AI models in whatever manner they see fit, we wish to bear witness to an AI-free mathematical practice. We also choose to do so as a community, so that this witness may have a shared, publicly visible form, as our contribution to cultural resilience and social ecology, and as an exercise of our academic and intellectual freedom.
Which indeed needs, like everything, a bit of exercise sometimes 🙂
Here’s what we mean:
- We will avoid from now on any use of AI (both in the form of general purpose LLMs and more specialized tools) for problem selection/solving, proof checking, literature synthesis/explanation, teaching aid, generative writing aid.
- The commitment is to a certain spirit: we do not outsource any of the intellectual and social acts that constitute mathematical understanding and practice, including routine ones.
- We value attention to detail, accurate crafting of arguments (be they novel and creative or standard ones), meticulous writing of text explanations, calm discussions.
- We try to form a community of people with similar concerns and taste. At the same time, we don’t want to convince anybody: we are open, not proselytizing.
- We are open to exchange and proposals. We are happy to let community spaces take shape only if the need arises naturally, without any expectation or forced engagement. If and when they do, those wanting to interact with other researchers without being annoyed by AI-reliance will find a social space that feels like home.
- We don’t ask anyone wanting to contribute ideas to stay AI-free outside the community interactions. We want to create a cordoned space, not to promote tech asceticism.
We see mathematics as part of human culture and the mathematical community as a vital component of the social ecology. There are many ways to contribute to both. We hope ours is a good one.