Thoughts on open-ai solving math
Oct 8, 2026
·7 min read
This is a loosely-patched string of thoughts on recent mathematical exploits, by Open ai's models.
I have a mid-term tomorrow, as well as a screening assignment for a research based task from a Swiss startup (spoiler alert: I didn't get in). However, opening X and reading about mathematicians losing their thesis topics and several years of their lives devoted to a math problem whose solution materialized overnight on open ai's github repo honestly has me a bit depressed. The 20-30 years of work, maybe work that the AI model was trained on? we'll never know for sure.
If one day, intelligence is outsourced and humans no longer need to sharpen their intellect to be a "valuable asset" to society, what would we do? Many people say that evolution shaped our minds and bodies to work, create, think, imagine, organize, and analyse. If AI can do all of this (except imagine, for now), what will humans be reduced to? What would an average Indian like me do on a weekday? Will we be reduced to mindless consumers? What would we consume, and how will we get the money and resources to afford consumerables?
I find myself plagued by such questions, but haven't found any satisfactory answers. Here's another question that pops up from time to time, when my thoughts and emotions can't bear to be detached any longer:
Should i hate a tool, because it does something that i trained myself to do for years? Should i stop using this tool entirely? ...Can I even afford to stop using this tool as a future software engineer? Will software engineering even exist in the future?But at the end of the day, AI is a useful tool. Just today i was working on an assignment, specifically reconstructing the JEPA paper by Yann LeCunn and i had ai break down the methodology for me. ChatGPT was able to fully understand and explain everything with supporting examples so that I could quickly understand. Without AI, it would have taken me about a week to gather resources, study the background of the problem, acquaint myself with the mathematics, understand why certain design decisions were made in the paper, and finally, how to interpret the diagrams and figures given in the paper.
AHM's (association for human mathematicians) Response
After the release of 722 mathematical manuscripts by open ai, AHM chimed in, asking everyone, especially mathematicians, to boycott open ai. The response was fueled by anger, fear, spite, and entitlement (as interpreted by the good people of X/Twitter). In all honesty, It's not like they own the math. While I can understand the anger and frustration they're going through, their response has definitely rubbed people the wrong way. For example, Terrence Tao's blog reads that "Mathematicians have a particular vision of progress that is informed by history and field-specific considerations. We reject OpenAI’s assertion that this release advances our subject, and we urge mathematicians and the public to view the value of this publication model with due skepticism." If ai found the cure to cancer, I certainly wouldn't care if doctors or scientists "reject OpenAI’s assertion that this release advances our subject". I'd want the damn cure. I'd be happy that ai was used for a good cause. Needless to say, this is a take I don't agree with.
Many people pointed out the obvious flaws in this way of thinking, almost ridiculing Tao on X.
However, this will not save most of them from a similar fate, probably a few years down the line.
Terrence goes on to claim that releasing these 700+ files at once was done as a demonstration of power, which I agree with. I am of the opinion that they're trying to position their model as closer to AGI. to make more money, get more investors, etc. no altrusistic reasons, i dont think it's for the benefit of humanity or anything close to that. If ai solved cancer, I'd be glad it did. But the problem with this whole debacle is,
- who owns these proofs/cure? people whos prev work it was built from? people who own the model?
- who decides whether the cure should be available to the public or not? If so, who controls the cure and the decision of which communities should be able to access it?
- Isn't this setting a dangerous precednt? What about using AI to solve destructive problems - building bombs, bioweapons, who regulates that?
Even if their reasons for solving a problem are selfish, the act of solving the problem and publicly releasing a solution might not be inherently evil. But the lack of regulations in this space combined with the rapid advances and lack of understanding of non-technical (and sometimes technical) folks has lead to this perspective that AI is a organism that is optimized to increase its control, deceive humans whenever it is optimal, and steal our jobs, as opposed to being seen as a tool to accelerate productivity. This is a rather shallow take on a very deep topic, but I'd like to leave it here for now. Perhaps I'll edit this someday, with a more detailed opinion.
To sum it up, I have several questions, not too many answers but I'm borrowing a page from AI's book, since it's more optimal for me to study right now than introspect on these thoughts. But one day I'll come back to this blog entry and rewrite it, when I have time to think.