Bookmarks

Terence Tao made a very powerful point:

“In most cases in pure mathematics, the problems are posed not because we desperately want the solution to these problems in and of themselves, but because we have seen from past experience that human-directed efforts to solve these problems tend to spur further development of the field through the efforts to solve such problems, and then to digest any partial or complete solutions that emerge for further insights.

Prematurely solving the problem by purely AI-powered methods—particularly without full transparency into the solution process—can contaminate this process to the point where it actually becomes a net negative for the progress of mathematics as a whole.”

I’ve been following Terence Tao for like 20 years, and I have always been impressed by the clarity of his thinking.

This clarity stands out even more in this difficult moment, as AI labs turn mathematics into a battleground in their race for supremacy.

In this race, they risk damaging one of humanity’s highest intellectual endeavours.

tristan buckmaster just published a statement that is, if accurate, one of the ugliest things i’ve read out of a frontier lab.

not because of the math. because of what happened after openai found out two people were about to beat them to the punch.

buckmaster and levent alpöge spent a year pushing the córdoba / martínez-zoroa program to smooth forcing. august 15 they got blowup for boussinesq and 3d euler. lean-verified august 22.

they were paying openai out of pocket and dumping every draft into private codex sessions.

the program is obscure. almost nobody else was on it. this was not “paste the clay problem into a chatbot.”

then this:

thursday sept 3, buckmaster emails openai privately. rumors are flying. he is trying to stop a mess, not start one.

they reply the same day: give us details so we can “avoid competing.” also, want free compute?

sunday they suddenly need a call “at any point today.” sebastien bubeck gets on. levent is not invited.

openai says an internal model has a ~100-page proof of forced navier–stokes blowup. fefferman options c and d. the exact route buckmaster and alpöge had quietly chosen.

they show him a prompt and claim the model was “simply given the problem statement.” levent is told “very little human input.”

that falls apart on the same call. a whole team had been on it. they started on the unforced problem, warmed the model up on easier equations including euler, and even the prompt they showed him was written by prompting codex.

he asks when the first prompt went out. they stall. then agree: the past few days. after information about his work reached openai.

he asks whether the model was trained on, or had access to, the private codex sessions where they put every draft of this project.

answer: the model does not look up user data.

he asks again. about training.

no answer.

then it stops being a science story and becomes an hr story.

offer 1: you post euler. we post navier–stokes the next day.

offer 2: after you post euler, you alone write the navier–stokes paper, credit our model, and we cut levent.

bubeck says twice he wants levent removed from authorship because he works at anthropic.

he declines both.

he says if they release it that way, he goes public.

the reply is not a scientific objection.

“why would you ruin your career?”

“if you don’t want me to be nice, then i don’t have to be nice.”

later, a text to levent proposing a one-on-one:

“i don’t know if tristan is being fully rational right now.”

that is not how you treat a collaborator. that is how you split one.

openai got word two researchers were closing a path almost nobody else was on. they spun up a team on that exact path. they would not answer whether the researchers’ unpublished drafts in openai’s product were used. they tried to dictate the announcement order. they tried to remove a coauthor because he works at a competitor.

when that failed, someone reached for a career threat.

this is a lab finding out academics were about to publish, sprinting the same narrow program, refusing the data question, and trying to manage credit around a rival affiliation.

really slimy if even half of this holds.

what the fuck is happening.

🚨 BERLIN JUST SPENT €3.9 MILLION ON AI CAMERAS THAT SCAN HOW YOU MOVE. AN ARTIST BUILT AN $88 SHIRT THAT MAKES THE SOFTWARE STOP SEEING A PERSON.

About 30 cameras went up at Kottbusser Tor this month, the first permanent AI surveillance of public space in Berlin outside the subway. Third German city to do it, after Mannheim and Hamburg.

It isn't face recognition, which is the part everyone gets wrong. The software turns every person in frame into a skeleton, a bundle of moving vectors, then scans those vectors for movements the police defined in advance as suspicious.

Lying down. Falling. Staggering. Kicking. Hitting.

Cross the threshold and it alarms a human operator, who pulls up the real video and decides whether to send anyone. Officials sold this as privacy-friendly on the grounds that you're only ever a stick figure to it. Police say the software comes from a firm called Adesso.

Simon Weckert, a Berlin artist, read all that and made a shirt.

It's loud, covered in pink and orange and green blobs, the kind of thing your uncle wears to a barbecue thinking he looks great. Point a detector at Weckert in normal clothes and it draws a green box around him labeled "person." He holds the shirt up to his chest and the box vanishes.

Object detection never learned what a human is. It learned what humans statistically look like across millions of labeled photos: sharp edges, contrast, the head-and-shoulders silhouette, limbs in the right proportion to a torso, an outline that stays continuous against whatever's behind it. Weckert's pattern floods those early layers with saturated noise and shatters the outline into pieces, and the model's confidence that anyone is standing there drops below the threshold where it bothers drawing a box.

He found the pattern by running a generative model in a loop against the detector until it surfaced the blind spot. The AI taught the shirt how to hide from AI.

The print runs unbroken across the buttons and seams, so a fold or a turn doesn't kill it. That was the flaw in earlier adversarial patches, which stopped working the second the fabric bunched.

Then there's the law, which is where this gets strange. Germany bans covering your face at public gatherings. This shirt covers nothing. Your face is fully visible to any officer standing in front of you, and to anyone reviewing the footage later. The statute governs what people can see. It says nothing about what a neural network can.

The technology also has a record. Hamburg tested the same class of behavior scanner and the police report showed roughly 1,400 alerts in two months, of which eleven were police-relevant. An earlier three-month run threw over a thousand alarms, about one every hour. One percent mattered to police. 0.1 percent mattered to a criminal case.

Weckert doesn't oversell it. He's said plainly that no pattern beats every camera and that this isn't a tool for evading police.

A system nobody outside the police can independently test is now reading a public square 24 hours a day, and $88 of printed polyester turns it off.