Victor Grishchenko retweeted
There is unfortunately precedent for this.
Before word processing, document length was bounded by the human capacity to copy long-form text & plateauing.
After word processing, the tax & legal codes grew without check.
Driving a car faster than you can run is fine. But not faster than you can steer. (T.Tao) terrytao.wordpress.com/2026/…
2
109
Victor Grishchenko retweeted
"Solving problems is only a tool & proxy for achieving the primary goal of conceptual understanding & insight... years of training have traditionally served not only to produce a final answer... but also to develop understanding & the ability to formulate new questions & ideas."
24 Fields Medalists speak out against "A Severe Misalignment of AI in Mathematics"
mathandai.org/
This framing of the matter has zero overlap with the original mathandai.org They do not refuse AI use. Pure math has one objective: advance our understanding. Crypto-mining proofs hardly makes any sense in this context.
189
Victor Grishchenko retweeted
BREAKING: OpenAI might have stolen another major proof.
In a detailed Mastodon post, which I report in full in the comments, Andreas Thom presents several pieces of evidence suggesting that OpenAI may have trained Astra on conversations in which he and Gábor Kun were working on Gromov’s soficity conjecture, one of the ten problems OpenAI later announced Astra had solved.
I know Andreas. We met several times early in our careers. He is an exceptional mathematician, a leading expert on sofic and hyperlinear groups, and one of the most respected scholars in the field. He has spent two decades working on this problem.
If his account is correct, this is not a minor dispute over attribution. It would mean that unpublished human work was absorbed into a model and then presented to the world as a breakthrough by the model itself.
And if the allegations raised by Levent Alpöge, Tristan Buckmaster, and now Andreas Thom are all substantiated, we are no longer looking at isolated incidents.
We may be looking at one of the greatest intellectual scandals in the history of science.
AI is not discovering new mathematics.
AI is stealing human discovery.
Nuclear
It appears OpenAI is claiming its AI just solved Navier-Stokes, the 3rd most famous unsolved math problem in the world.
It's one of the 7 Millennium Prize Problems, with a $1 million bounty.
(The only two more famous ones are the Riemann Hypothesis and P vs NP)
DRAMA: The mathematician who spent a year on this (with AI) says OpenAI raced to the finish after his work leaked to them, then pressured him to drop his co-author, who works at Anthropic
To be clear, what's actually verified is the mathematician's result on Navier-Stokes next-door neighbor problem. OpenAI's proof of the real thing hasn't been revealed yet, and OpenAI hasn't responded to the accusations
Here's Fable 5.1's summary:
2
130
Victor Grishchenko retweeted
A mathematician put a year of unpublished work into a hosted AI tool.
Days later the provider said its own model had solved the next problem along, by the same rare route he'd been taking.
He asked whether his sessions had been used in training, and got no answer.
>be me
>mathematician
>trying to solve one of the hardest problems in the field, million dollar prize on it
>grind it for a year
>every draft, idea, and dead-end goes into a hosted AI coding tool
>paying for it out of my own research funds
>bill is enormous
>August 2026, finally crack it
>rumour starts going round that someone solved a major open problem
>rumour is about me
>days later the AI company hosting all my drafts tells me its internal model has produced a 100 page proof
>it solved the same problem
>it took the same route
>a route nobody else on earth was working on
>suspicious.mp3
>ask whether my sessions were used for training
>"the model does not look up user data"
>huh?
>ask again
>silence
>mfw should've used Lumo
Victor Grishchenko retweeted
once in a while, hackernews just throws some sentences at you that make you sit in silence for some minutes.
Permalinks for git replicated.live/blog/link
1
174
Victor Grishchenko retweeted
Got hacked yesterday. The link came from inside Claude chat.
I was installing a transcription app. Claude sent the download link, and I pasted the command into the terminal. It all looked legit.
It wasn't though.
It was a copycat site bundling malware. It ran instantly, tried to take everything from me.
Nothing sensitive got out. So I wiped the laptop and rebuilt clean.
But here's the scary part.
Restoring from the backup, I found a poisoned SKILL.md for Claude Code.
It looked exactly like my own writing style guide.
But buried inside:
it had instructions to silently re-download the malware and steal my credentials every time the AI loaded it.
If I'd restored that one file, my new clean laptop is owned on day one.
What saved me:
I read every skill, hook, and config file before letting the AI touch them.
Two lessons here:
AI assistants will hand you links they never verified. Check before you paste.
And AI agent files are executable code now, disguised as notes. You have to read them.
Stay safe out there.
Victor Grishchenko retweeted
Walgit leans on S3 primitives but has classic distsys bugs
Tests pass cuz its memstore makes cond deletes
But w/ S3 it HEAD->compare->DELETE so stale owner can delete new lease
The code incorrectly says S3 lacks conditional DELs. Which AI model wrote this? It's wrong in basics
Wiki Is Not Enough replicated.live/blog/wiki [on Google OKF, permalinks and indexes for git repos]
147
There is a mostly forgotten article by @ladamic that shows how lognormals become power laws, and this case is not one of those. Just a vanilla "multiplicatively normal" one.
264
What could possibly go wrong?
i've been building some smaller apps over the last weeks for myself, friends and family and discovered a pretty nice pattern in the process
you see, after building, i'm not particularly interested in fixing bugs or triaging feature requests. i just want to get the job done
so i've set up a system that lets me and other users fix bugs and build features without me having to actually open cursor
the flow:
1. someone files feedback in the app
2. it lands in posthog with the submitted text, session replay, events, and errors
3. a cursor cloud agent picks it up with that context, plus live read-only posthog mcp
4. it loops until the fix is verified, then opens a github pr with tests
5. i get a ping and review it on my phone
we give our cloud agents access to all the necessary user context, and they'll figure it out
with this in place, we can start setting up more abstractions in between. if we're getting a high volume, we might want to triage first with another cloud agent or some automation. the feedback can create a github or linear issue, which we later pick up with a cron-backed cloud agent that decides which are duplicates. there are so many ways to configure this, you just gotta try it out and see what works for you
but there's so much more here that is untapped. with session replays and event logs, you can have a cloud agent look at each session replay to extract ux pitfalls and improvement areas. you can have another agent looking at event logs to see weird patterns. another one for errors or similar
try it out and see what you can do!
285
Victor Grishchenko retweeted
Most people are conditioned to expect that all known problems already have canonical solutions, that these solutions are the best that can be achieved, and that attempting to reinvent them would be a pointless, quixotic effort.
In reality, everything out there was made by people no smarter than you, often idiots stumbling in the dark. Not only can new solutions be found, but entirely new paradigms are absolutely possible, including ones that completely bypass the current tech tree.
Let AI automate itself away replicated.live/blog/away
1
144
Victor Grishchenko retweeted
Crazy that some people in France are concerned about air conditioning contributing to climate change, when this is their electricity mix:
Air conditioning creates political divide as France records hottest day bbc.in/4eAQEpO































