Why AI is bad for mathematics

updated on November 14th, 2025 at 10:46 am

I was recently struck by the use of AI in one of my first interests: mathematics. I wrote a letter to another mathematician professing my disdain over this development. I did acknowledge that if we want to maximize the production of truth, then developing advanced AI systems capable of mathematics is surely the way to go.

But do we really want to maximize the production of truth? Mathematics serves more than just a source of logical truth. When Euclid wrote of the bisection of a line, we don’t just marvel at the beauty of the construction, but of the elegance of the human mind which obtained it. When Gauss proved the fundamental theorem of algebra, we don’t just find a reason behind the existence of complex roots of nonconstant polynomials, we feel the passion behind the search for solutions. This admiration for the art of mathematics goes beyond the production of truth. Instead, this admiration is of the journey for truth.

Some have made the point that mathematicians have used machines since the days of simple calculators. Yet, these calculators are mechanical in nature. Even modern computer algebra systems such as Sage still only operate in the mechanical domain. With modern AI systems, for the first time, we are seeing the nascent stages of a machine that can suggest creative lines of thought, and that is fundamentally destructive.

For, by embarking on the journey to automate mathematics on the creative level, we mechanize the search for truth into an industrial production. We take away the important of the human element. Now, every student going into mathematics will no longer look at mathematics as an intellectual journey to be shared by fellow travelers, but as the process of assisting the machine to find truth. But for what?

When I was a graduate student, myself and other students often had difficulty with understanding every step in various proofs we had read. Sometimes people would come to me and ask me why something or other was true. It was a great joy to work out the details and explain it to them, or have such details explained ot me. I think it won’t be long now before an advanced AI will be able to do the same, and because humans often take the path of least resistance, students of the future will interact with AI instead. The sense of community will dissipate, and again, the emphasis is shifted to utilitarian production. Even if we gain some benefit from the truth out of AI, what is the purpose of life on this planet if we become subservient cogs to the machine?

This phenomenon isn’t restricted to mathematics. In fact, it’s not really just an AI problem either, as much as I hate AI. At some point, we decided that the only thing that really matters is economic production, fueled by the human instinct for greed. AI is just the apex technology in that process. We still have the freedom to pursue other objectives, but only in the increasingly restricted realm of objectives that don’t affect economic progress. And, the latest in the long line of such restrictions is the discarding of a sense of community to further AI-assisted production.

Some will consider this tradeoff to be worthwhile. I consider it offensive, and I think it should be stopped.

(This post is the script for an upcoming video on my YouTube channel.)


My website does not have a commenting feature. Instead, if you like you can use this form or send me an email at radioactive (at) jpolak (dot) org.