The noise. The slop. The conversation dominated by AI stuff. I guess that’s a welcomed break from mulling over the fact that we are all dying, if not quickly cooked then slowly poisoned1. Meanwhile our institutions want us to answer the following question: “Is AI killing math?” First no, and then what a stupid question to ask, suggesting that it’s an option. Matlab didn’t kill linear algebra and only ignorance can hurt math: it is wherever you are willing to see it.
What some AI will end up killing though, is us all and faster than expected2. And maybe slightly before that, teaching bots would have replaced us: They politely answer any question without judging, start class on time and would probably train our students much better at acing those tests3 that we sadly tag as mathematics.
Indeed OpenAI solved some important4 open problems, failing at thousands of others and resorting to theft. That’s by design, right? Let’s be honest, we had it coming: we have collectively been passing overly re-heated ideas as new for as long as I can remember. Spray-painting a coat of new and original to be able to publish in a better journal seems to be in the job description. Our lazy ignorance rewards quantity over quality and makes us treat math as a competition instead of a collective art work.
The other day I was explaining my job to my teenage daughter. The teaching part she knows, she has teachers. The rest I told her -and mostly myself- is writing stories about imaginary spaces, its inhabitants and their geodesics, and try to publish them in journals that nobody really buys or reads. Back in the days those would get printed to be displayed in libraries and reprints were passed around like candy. I find interesting stories to build on, get people hooked by my favourite characters, and follow friends in pursuit of exciting adventures to be recorded in those journals that nobody even prints anymore, but that universities still use as token of excellence. I do mourn the excitement of having my work exposed and running my fingers on the glossy finish of the intensely messy experience that is to write a math paper.
But what I mourn even more are the days when climate change was a distant threat that a few laws would fix. The tipping point has clearly passed, but data centres will continue to grow whether we use them or not, and we are still cooking. Now what?
We can still dream of a beautiful future5. What if locally and sustainably powered AI allowed us all to become super-versions of ourselves? We would be able to keep the carbon underground and restore biodiversity, we would all have the job we love, eradicated drugs and crime. Every single human would be a well adapted yet deeply unique perfectly fitting piece of the beautiful puzzle of a flourishing earth with clean water, clean air and delicious food. We wouldn’t have to referee badly written papers, in fact we would just share ideas in front of a blackboard, without petty attribution fights and only in pursuit of knowledge. Meanwhile an AI would politely take notes, diligently fill in the gaps and upload a summary checked by lean in our extra memory chip while we dance ceilidh6.
The fact is, AI is here to stay, Nash equilibrium predicts that we (as pawns in our current patriarcapitalistic world) will use it and that dissenters will disappear7. There is no opinion to have, merely a few predictions to make.
Of course a few improvements are still needed for an AI to provide us with clean water or clean air, and right now it’s rather set on the trajectory to kill us all. The sustainable Swiss AI Euria is still quite a bit weaker in math than my human colleagues, and if ChatGPT is by now more knowledgeable than them, I am not (yet) desperate enough to tolerate the lame mansplaining that systematically ends by asking for money. Not unlike amphetamines, great on a short deadline but will turn your brain into a mushy financial sinkhole if you start liking it a bit too much. So I will try to stick to my human peers for a bit longer. That is, until whatever kills us all, or the next grant application, whichever comes first.8
- See the post How dare you, and Vadim to answer your question our excuses are that once we figured out that planes keep flying whether we are in them or not, we decided that there’s nothing we can do. ↩︎
- With some (completely made-up) 10% probability according to some AI bigwig, and likely following one of the thousands scenarios that is already laid out somewhere in our gigantic collection of stories. ↩︎
- Tests to land a bullshit job that the world would be better off without, and that said AI would be better at doing anyways… Remember, we’re suffocating, we need clean air and clean water. ↩︎
- That is, tagged as important by a subset of the math community, usually consisting of middle aged white and wealthy winners of our patriarcapitalistic system and mostly based on what the biggest alpha males and their army of students have been failing at solving. I can’t help but wonder how important those look to the people whose water supply is contaminated by a data center. ↩︎
- And more urgently, get out learn how to grow food or any other survival skill. ↩︎
- Yes, last month I was at a conference in Glasgow, an all blackboard one how I enjoy the very imperfect stories we try to share and our flaws in doing so. ↩︎
- See “The AI assisted mathematician’s dilemma”. ↩︎
- Reader please forgive my gloom, close your computer and find a tree to hug, life will feel beautiful again. ↩︎
Received 31 August 2026, revised 17 September 2026.
Add to the discussion