SERIES · AI is making papers cheap to produce but they remain costly to referee. Proofs and Prompts is publishing a series of posts from 26 October on the future of publishing, papers and credit. Questions such as how to disclose AI use, who bears the cost of checking, what authorship means and what papers of the future might look like will be addressed. This post is part of the series. Each is written independently and reflects only its authors’ views. All posts in the series
I am about to graduate with a math PhD, with zero published papers, and a subpar thesis that’s just barely enough for me to graduate.
I have no illusions that I am on the same intellectual tier as the tenured professors, postdocs, and PhD students much more successful than me writing on this blog. I am no Yitang Zhang. I readily admit that I am a low-level mathematician compared to even the peers in my department. And much of this is my own fault. Significant stretches of my time in the program were spent procrastinating and being distracted, not to mention just plain being stuck on a project because I wasn’t good enough.
Realistically, I will not, and should not, be getting a research postdoc. I will end up in an overloaded teaching position or working a regular 9-to-5 job. And I do not have contempt for this future. I am sure I will manage to find a job that is fulfilling for me.
Under the old rules of academia, my life in mathematical research was over. I was supposed to quietly pack up my stuff, and leave research mathematics to the REAL mathematicians.
But I am still interested in doing math. I still think my research direction has genuine merit, and I still feel like I have more mathematics to give to the world. Thanks to AI, it is now feasible for someone like me to keep going. With language models accelerating the time-consuming parts of research, it is realistic for someone carrying a heavy teaching load or working a day job to produce novel, interesting mathematics in the little free time they have.
That is what I am going to try to do. And what is wrong with that?
I wanted to write this down because when I read essays here, whether pro-AI or anti-AI, people like me aren’t considered a part of the conversation.
We have an opportunity right now to ‘catch’ the people who have fallen by the wayside. People who just barely survived their PhD program and still have good ideas. Or people with just a Master’s or a Bachelor’s in mathematics who didn’t want to go through the grueling 5+ years of low pay for a PhD. Or those that went through years of postdoc positions and failed to secure a tenure-track job. We, the people outside of research academia that still wish to continue with mathematical research, exist!
But almost every conversation among established researchers seems to forget that. This is understandable because, before AI, many of us could not feasibly do serious research mathematics outside of a research position. Research math is incredibly time-consuming. Even if someone had a novel mathematical insight, keeping up with the latest literature, checking citations, and writing things up in an understandable, presentable way was a time-consuming process. It was not feasible to do without a lengthy dedicated research time in their schedule, afforded by a research position. We were not considered “real mathematicians” for a good, understandable reason. (There are exceptions, of course. Those outside of research jobs have produced a great deal of amazing research before AI. But we must admit that it is now MUCH EASIER to do research math outside of an academic position.)
Now, with the advent of AI, it IS realistic for us to continue participating in mathematical research.
So could we be brought back into the conversation?
Because right now, when I read through the posts on this blog, people like us are completely ignored:
- I see reports from elite gatherings like the Harvard CMSA summit on Math and AI, or calls for societies to set norms. Will those boards and summits ever include people like me? Will they care about the opinions of an overloaded teaching faculty member just wanting to continue exploring the direction of their thesis? Or the person doing math research in their free time after their 9-to-5? I am afraid that these organizations will just be another society of already successful mathematicians writing rules for themselves.
- There is some panic that AI will create a torrent of papers and there simply won’t be enough qualified people to peer-review them. Really? There aren’t enough people to review them? Thousands of math PhDs graduate every year without continuing a research career. Is there really a shortage of capable people, or is our definition of “capable” severely restricted? With AI, it is now much easier for those without research positions to keep up with the latest developments in their fields. Would that not drastically INCREASE the number of qualified peer reviewers?
- Several pro-AI posts suggest that because AI makes writing papers too easy, we should stop writing papers or devalue them in favor of live seminar talks and in-person workshops. Do people realize how exclusionary this is? It is a luxury that can only be afforded by those with grants and dedicated time off for travel.
- I see professors warning that because of AI, we might “lose an entire generation of mathematicians”, while others decry that AI will reduce math from a profession to a mere “hobby” done in people’s spare time. Where was this panic when the academic job market was throwing away the vast majority of every graduating PhD class for decades? And why is doing math in your spare time, for zero money and zero career credit, treated with such contempt?
I have lost the game of academic research careers, and I have made peace with that fact. What frustrates me is that the mathematical research establishment pretend I don’t exist. When a tool finally comes along that lets me keep doing the thing I love, making small but real contributions to my field from outside of a research position, I continue to be ignored. Instead, one side is telling me that my work doesn’t count for using AI, while the other side talks only about how it can be directed to better serve and preserve the scant few dedicated research positions.
It is telling that even in a fictional utopian vision of mathematics in 2038, set in a literal post-scarcity world, one can only imagine the mathematical community through the eyes of... an ETH Zurich professor and a postdoc heading to a three-month retreat in Princeton. Wow. Just wow. Even in the wildest, most utopian fantasies of the future, people like me don’t exist.
Received 1 November 2026.
Add to the discussion