upvote
Terry Tao has actually been one of the more prominent voices in the math community exploring AI for cutting edge mathematical discovery. This particular post is a bit softer but he has also written a lot about using AI assistance for serious core research

Nov 2025: https://terrytao.wordpress.com/tag/artificial-intelligence/

https://academy.openai.com/public/blogs/terence-tao-ai-is-re...

reply
What makes this a hobby project? He’s a university professor so developing teaching material is part of his job.
reply
> always a hobby project, never something serious.

I don’t know what you’re reading, but always and never are strong words. I’ll predict by this time next year you’ll have seen some pretty serious AI uses, and can no longer say always/never. Widespread use of AI coding is brand new, and the models only just barely got good enough to do serious things. It’s way too early to be using words like always and never, but FWIW I’ve already seen some serious uses. There are good reasons personal blog posts rarely talk about ‘serious’ production code; it may be against organizational policy, it may involve code that isn’t’ public, it may reveal proprietary information, and more…

reply
deleted
reply
Having started using Claude Code at work, I think coding as we now know it will probably no longer be a career path in 5 years at the outside.

I’m old. If I had to, I could retire tomorrow, albeit on a restricted budget. But I worry about the younger folks (like my 25-year-old nephew) who haven’t built up the resources to survive without working who are in the field right now. There’s going to be a mega disruption and writing code is going to go the way of calculating square roots by hand or hot metal typesetting. There will still people doing it, but it will very much be a niche endeavor.

reply
It's a race to retool. If he can retool to an agentic coding environment quickly he will be OK. Might have to build up the skills outside of work if the current job doesn't allow for it, but I'm less worried about younger people realigning quickly. They're used to it, and younger people are so much more resilient/used to disappointment than we are (economic prospects have been so fucked for Zoomers for so long by so many "once in a lifetime" events and they still soldier on which is a huge testament to them).

Whether reviewing agentic code rather than writing it is a job he wants to do... Different question.

Its older people who can't or won't retool that are going to find that in the game of musical layoff chairs they won't have a chair left when the music stops (and some I've met haven't really internalized that they're even part of the game...)

reply
But he's also using AI for formally verified math and for ideas in solving math problems. The part about it being ok because it is a supplement just means ok that these aren't formally verified and may have bugs, and may also mean ok to not credit the AI for the paper as it is just a visual supplement and not the main work.
reply
deleted
reply
That is how it starts, trust is built on hobby projects.
reply
What do you mean not serious? He’s developing visual aids to teach students and to accompany his mathematical research papers. Also, not in this post, he’s been actively using LLMs to do real math research, that is, to prove theorems and solve problems.

Teaching, research and publication are the core activities of his job as a math professor. How does it get more serious than this?

reply
Serious things tend to be long and tedious and potentially full of proprietary information.
reply
Open science has plenty of long and tedious articles.
reply