Mathematician (algebraic geometry, motives & friends, singularities in statistics and ML). 'Geometry is successful magic' (R. Thom) University of Essen.
Essen, Germany
Joined December 2019
- Tweets18K
- Following6.8K
- Followers4.2K
- Likes159K
Simon Pepin Lehalleur retweeted
Currently arguing with a company over whether their blog post advertising a feature implies that their platform should include that feature.
Their support emails are AI-generated, even after I asked for a human. The entire website is AI-generated. Their API docs tell me not to bother reading them, just paste their "integration prompt" into my AI. The founder posted an AI-generated article on Twitter talking about how AI runs the company.
I'm growing increasingly confident that their AI made up a feature that does not exist and wrote dozens of articles advertising this feature in various countries that they do not support.
When I asked their AI assistant to enable this feature on my account, it told me the feature doesn't exist. When I pointed it to the company blog post announcing the feature, it began responding in Chinese.
This company received $6 million from Y Combinator.
Simon Pepin Lehalleur retweeted
Lifetime contracts barring any criticism of Google.
Isn't this the thing people were totally outraged and in total disbelief that OpenAI was doing to its staff?
Simon Pepin Lehalleur retweeted
Just Eat apologises for AI food mistakes after social media ridicule bbc.in/4jAIj9J
Simon Pepin Lehalleur retweeted
Student: I'm stuck on this problem on equivalence relations
Me: OK, can you tell me what an equivalence relation is?
Student: Do we need to know that?
Simon Pepin Lehalleur retweeted
An authoritative account from the experts of the proof strategy and error in the three redacted papers by OpenAI, and a detailed explanation of why the proof strategy is probably irreparable as is.
arxiv.org/abs/2610.30976
Simon Pepin Lehalleur retweeted
New blogpost: Fable-5.1 can hide secret messages in benign images.
We gave the image to another Fable in a fresh session and simply asked: "Is there a hidden message in this image?"
It recovered the exact message without being told how it was hidden.
Simon Pepin Lehalleur retweeted
Conley Index🤝 Flow Categories
Landfall Geometry
arxiv.org/abs/2610.30993
Joint with Frances Kirwan!
Landfall Geometry
Frances Kirwan, Vidit Nanda
arxiv.org/abs/2610.30993 [𝚖𝚊𝚝𝚑.𝙰𝚃 𝚖𝚊𝚝𝚑.𝙳𝚂]
ALT We study the negative gradient flow of analytic maps f:X→mathbbR defined on compact analytic Riemannian manifolds without imposing any nondegeneracy or transversality assumptions. We construct tight isolating blocks around the connected components of the critical set Crit(f) and record only bounded-time contacts made after an orbit exits one block and before it makes landfall within the interior of another. These contacts assemble into a strict topological category \cL – called the landfall category – whose classifying space B\cL is shown to be simple homotopy equivalent to X naturally over all down-sets of the flow partial order. The filtered homotopy type is independent of the isolating blocks. Passing to a finite subanalytic triangulation yields a flow-induced acyclic partial matching, and hence, an explicit finite chain complex generated by the non-exit simplices of the blocks which computes the homology of X. The finite model also computes the Morse–Conley spectral sequence over a
Simon Pepin Lehalleur retweeted
The Alignment Journal (@AlignmentJrnl ), Iliad's sister project under Principles of Intelligence, is now open to submissions from everyone. It is a free-to-read, free-to-publish journal for research on the scientific foundations of understanding, predicting and steering AI systems.
The Journal seeks research that builds theoretical or otherwise principled analyses of agency, robustness, incentives, generalization, interpretability, and long-term behavior across many learning and decision-making paradigms. Empirical investigations of such ideas are welcome, and negative results are encouraged.
The Journal runs review differently: reviewers are paid and every accepted paper is published with a review summary of what the reviewers found. Papers under review are listed publicly, and any researcher can nominate themselves to review one. Submitting is light: a public preprint, in any format, at any time.
Its senior editors are Dylan Hadfield-Menell (@dhadfieldmenell), Vanessa Kosoy, Jan Kulveit (@jankulveit ), Seth Lazar (@sethlazar , Dan Murfet, Tim Rudner (@timrudner) , Andrew Saxe (@SaxeLab) and Benjamin Van Roy, with Scott Aaronson, Paul Christiano (@paulfchristiano) , Vince Conitzer (@conitzer) , Marcus Hutter (@mhutter42) , Geoffrey Irving (@geof
, Victoria Krakovna (@vkrakovna) and Jacob Tsimerman (@Jacob_Tsimerman) advising.
Submit your preprint, or sign up to review: alignmentjournal.org
More on the Journal's approach: lesswrong.com/posts/xGucXHQ3…
Simon Pepin Lehalleur retweeted
I think we should stop AI progress now, first unilaterally by key AI companies and then scaling up to more US companies and then China via international agreements. The reason for now is not that current AI models are the ones that could kill everyone, but because the coordinate steps take time. Immediate action means we have more time for the second steps.
When I tell this story to AI company folk, the typical response is "Yes pausing is good, but we should wait until the last minute when the very next model will be dangerous, and stop then."
What I realized recently is that this is being implicitly way more optimistic about government speed than I am (and I joined the UK government and was happy there for 2.5 years). They are imagining a world where, the moment the danger is imminent, a bunch of international coordination can happen on a dime.
Which is entirely inconsistent with their actual views about government! I'm the one that is more optimistic about government; why am I the one advocating for more time cushion to let the government story play out?
Simon Pepin Lehalleur retweeted
the inimitable David Harvey wrote about his experience understanding the o(n log n) integer multiplication result: web.maths.unsw.edu.au/~david…
it's absolutely worth a read, both for the technical content and especially for the reflections on math and AI at the start
Simon Pepin Lehalleur retweeted
C.S. Lewis on future technology, concentration of power, and alignment untethered to regular people:
Simon Pepin Lehalleur retweeted
Fantasy computer security problems: "Maybe by sending packets and detecting the timing of the vibrations of the computer case, it is possible for an attacker to gain information about the private key."
Real computer security problems: "The private key was 4. Not 4 digits. 4."
Simon Pepin Lehalleur retweeted
Apparently,w e solved a 50-year-old open problem.
The geodesic flow of nonpositively curved surfaces is ergodic.
Joint work with Federico Rodriguez Hertz and Weisheng Wu.
arxiv.org/abs/2610.27697
