A thread that is probably misguided. Let's call it: An ordinary language response to a structuralist problem in philosophy of mathematics. It's not very catchy.

He mobilizes some truly incoherent arguments to defend the position that classical logic is in some sense the One True Logic. He also epitomizes the stupidity of analytic metaphysics. Yet another second-rate modal logician who turns to philosophy to justify his useless slop.

His views on logic, ontology and metaphysics are all very bad. I also didn’t like the way he engaged with Thomasson

Those few of you for whom this is applicable are likely already aware, but remember this is in a few hours. The corrected link is sci.brooklyn.cuny.edu/~noson…. I've been posting sturm und drang because that's fun in a way. But math is still alive and active as ever.

Tomorrow in my seminar on set theory with urelements, we’ll discuss philosophically motivated axioms governing urelements. We’ll look at Plenitude—the axiom that every ordinal is equinumerous with a set of urelements—as well as its stronger version.

the math hate churning out of this shithole is really something ...

Where are the philosophers/sociologists of mathematical practice in all this AI x Math ruckus? The conditions are never better for them to put their research in application and get involved practically. So far it seems like only mathematicians are putting their voices out there. Are they just not as vocal/online as the mathematicians (or more cautious to speak up because we're now beyond theorizing, when actual livelihood is at stake)? Or is what they have to say still under review at journals?

many important problems are not "long-standing", and many long-standing problems are not important :)

not a single logician, I love being a second-class citizen within mathematics openai.com/index/advisory-gr…

Wtf CIC + LEM proves Con(ZF) ?!? Which then gives a proof of Con(ZF) in axiom-free Lean + LEM. Found by Mario Carneiro with Fable. Finally some progress on the big questions in type theoretic metamathematics...

Does the air at OpenAI taste different from what we breathe?? Or the water maybe?!? Why are people who “work” there so out of touch??! From that 20-something intern who said “math is already a mostly solved problem, there will be bigger fish to fry” to my idol saying “we have to slow down AI. The pace is insane and there's no reason to be this fast, no reason at all.” God!! The arrogance and disconnect is wild!!

A whole new generation of LLM-generated Ph.D. applicant emails is in my inbox this morning. For the love of god, if you are going to forfeit your voice, please don't cede it to Claude. I would never hire Claude

In general I feel that Gowers has lost the romance of mathematics. I do mathematics, as do most people I know, to surprise myself. Finding things that catch me offguard ("Oh, I never thought about it like this"). That's just so much more exciting than this dull ambition.

Duh ... its the set of ordered pairs whose first element is an AI and whose second is a Math ... whaddathey teachin kids at Harvard these days

The soothsayers guild strongly opposed the invention of meteorology, but most of the public shrugged and said they should have seen it coming.

If the goal is human understanding, I think there are some concrete steps that can move us closer to that goal. 1. Make expository writing count in job applications 2. Make viable an educator career path for your students. Educator broadly construed: classroom teachers, science journalist, publishing/media business etc. Provide training, host workshops, introduce connections 3. Work with philosophy, history, and anthropology departments to further the study of mathematical practice 4. START PAYING YOUR TEACHERS & LECTURERS MORE It's simply not a sustainable narrative to champion for human understanding while driving its stewards to bankcruptcy.

Find someone that loves you as much as economists love commenting on mathematics. 😄

The disempowerment is coming from within the house. Not by the AI tools themselves, but by the companies building them that insist on building AI tools meant to replace rather than collaborate with humans

how about this? whenever a big AI lab solves a Millennium Prize Problem, they pick a major open problem (something that still seems well out of reach even with modern LLMs) start an endowment with a $1 billion prize for it, call them the Billennium Prize Problems

The future I want is 10K humans and AI agents in a heterogeneous mix with complementary backgrounds and skills all collaborating together to advance human knowledge, giving appropriate credit to the right humans wherever it is due. But AI companies don't want the humans

"SOP2=SOP3", arxiv.org/abs/2608.13291 This surprisingly short argument answers an old question of Džamonja and Shelah from 2004, and seems to be the first significant AI-assisted advance in mathematical logic. It is also a result we were lucky not to have found back in the day.