September 9, 2026
(Update to my August 23 statement below) My enthusiasm to use AI to work on safety problems lasted exactly 17 days. I leave such things to more clever people than me. The Association of Human Mathematics was launched today. I hope to learn about and contribute to issues such as teaching practices, journal policies, and mentoring the next generation of mathematicians through it.
August 23, 2026
This is an update to my original August 18 statement (found below). This update is a result of repeated exchanges with friends, colleagues and mentors Eren C. Kızıldağ, Abhishek Dhawan, Cheng Mao and Partha S. Dey, and my own ongoing education from different resources. I attribute my earlier statement partially to my own ignorance, and driven by mathematical romanticism -- a luxury while Fields Medalists can afford, is perhaps less pragmatic for those of us who are struggling to make out our existence in this ever-changing world.
I am unsure about my exact stance at this point with respect to capable AI. Following Levine's characterisation of mathematicians as artisanal, industrial and civic, I think my current position is somewhere between artisanal and civic. In my own academic research work, I remain artisanal, not using AI myself, while also not abandoning colleagues and coauthors who use AI with proper disclosure in accordance with the Leiden declaration. Simultaneously, I am increasingly becoming passionate about harnessing the acceleration that AI can produce (its trial and error capabilities have become unreachable humanly) to study problems on AI safety, with a personal goal to educate myself on what prevents omnicidal futures for humanity.
On this page, I will continue recording my own thoughts as we pass over this critical period.
Statement: August 18, 2026
Every mathematician has their own personal reasons to practice the discipline. Being a first generation academic, doing research mathematics to earn my bread was never a natural career choice for me. My initial reasons to take up mathematics were inherently selfish. I first embraced it mostly as an escape venue from life's troubles.
My reasoning was somewhat like this. While the decisions that we take in our lives may not always result in the outcome we were hoping for, this is never the case in the world of mathematics. If a mathematical result is true, it is true, and if one makes the right approach, one will reach it. In other words, I am completely in charge of my own fate inside the mathematical world; if I make the right choices, I will be rewarded. A utopia.
Navigating this utopia, appreciating existing results, and searching for new ones, has saved my life multiple times in different contexts. The realisation that a result might be true and the sheer beauty of the result driving me to commit a portion of my own finite time in this world to internalise why is it true (or why not), constitutes a journey, at the end of which I often came out empty handed in terms of my original goal. However, having the privilege to undertake this journey has always been the sole appeal of mathematics for me.
In terms of mathematics, this journey has always rewarded me with new insights and perspectives on the specific topic. Sometimes I found new definitions and ideas that steered new explorations. Beyond such rewards, making this journey has at times provided me with a sense of purpose when I failed to find purpose in my life, helped me with my mental health, strengthening my resolve to look and move forward in life, and forged deep friendships with some brilliant minds, with whom I was privileged to walk together. I always came out a better man at the end of such a journey.
From my limited experience in experimenting with large language models (LLMs), to me it seems the purpose of such technology is to precisely eliminate or cut short this journey. While this may be exciting to result-driven mathematicians, this kills a core component of why I choose to practice the discipline. I care very less about what comes at the end of such a journey, I only want to keep on undertaking it. Thus, for me, it makes no sense to use such technology in my own work. In particular, using them also feels like insulting the many instances when research helped me to get my life back together. I thus choose to stay away from them.
-- neeladri maitra