AI in science: recommended content
Due to the increasing skill of AI models, current times feel turbulent to anyone whose job it is to think deeply. I elaborate more on the problems AI poses for academic science in this post, but here I want to provide a summary of what I believe to be worthwhile discussions or analyses on the topic.
There are many different interpretations, opinions, and expectations for the impact and role of AI in science, now and in the future. I hope this diversity is well-represented in the content posted below. I present these as-is, without judgement or endorsement.
These posts are almost exclusively long-form content (podcasts, interviews, video essays, long-reads, papers), which I believe is really the only way to engage with this topic, apart from discussing with others or experimenting with the technology yourself. Best to avoid the noise on social media, which is full of alarmism and bad, misinformed takes.
For historical sake, I will move outdated entries to an archived section rather than delete them. The rest I try to list somewhat chronologically (newer entries appended below).
AI in science
- Tristan Buckmaster’s nuanced interview on Numberphile, which goes in-depth on controversy around OpenAI’s Navier Stokes MP-solution (see Buckmaster’s publications just prior to OpenAI’s)
- A Severe Misalignment of AI in Mathematics, open letter by Field medalists
- 100 reactions to 100 solutions: a snapshot of the initial responses to OpenAI’s dump of hundreds of solutions to mathematical problems
- 3Blue1Brown’s case for optimism in the field of mathematics (and perhaps science more broadly)
- A Google study on “early insights” of AI in science
- TheoPhysEval shows GPT-6-Astra and Opus 5.5 are starting to work through difficult theoretical physics problems with ease
- How to avoid AI brain rot, discussion of how to use AI in a way that prevents skill atrophy and promotes learning
- Early career researcher responses to AI in math, including the appendices
AI in broader society
- Bill Gates has written a long blog post where he expresses concerns and potential solutions to AI in society.
- In an interview by Ezra Klein, Bill Gates aggitatedly defends his case under pushback from the Ezra
- Andrew Ng interview by The Economist (paywall) - pushback on doomism
- Ethics forum ETH on free speech and AI surveillance
- WaitButWhy’s two-part series on AI revolution from 2015 is still worth a (re)-read in 2026