Feeling like you're on the verge of a major scientific breakthrough after talking to AI for hours on end would normally be considered a warning sign that you're spiraling into an AI-induced psychosis. But for PhD students its considered a normal working day. www.theguardian.com/technology/a...

Black Box: The Chatbots | Spirals | Ep 1 – podcast

Across the world, hundreds of people have come to believe they have made extraordinary scientific discoveries with AI chatbots such as ChatGPT, Claude and Gemini. Others say their AI has ‘awakened’, o...

theguardian.com

Me waiting when OpenAI will finally contact me and ask to shepherd the release of their next big quantum breakthrough.

OpenAI: Hey, we just used AI to disprove ETH! Me: Hey, I just used AI to produce a PDF that doesn't solve my much simpler problem, but at least it suggests some promising-looking next steps that will produce another similar PDF!

This is how you properly acknowledge that you used AI to write your paper.

Gautam Kamath@gautamkamath.com · 5d ago

How do you tell if authors fabricated their research results? Sometimes: they just tell you. Read, e.g., this caption carefully. Another banger of a post by my friend Nihar B. Shah on his blog, where he found at least 20+ such #ICLR2027 submissions: researchonresearch.blog/2026/10/20/s...

Every paper should include illustrations like this. www.dropbox.com/scl/fi/zmqzd...

Current state: "deep in thought with xhigh effort".

Wait, is he talking about me? www.youtube.com/watch?v=lXrD...

argon doesn't trust me 😭

BREAKING: Quantum computers can add 5-bit numbers ~70% of the time (when discarding runs that totally fail). arxiv.org/abs/2610.14202

Below-threshold execution of fault-tolerant ripple-carry addition on a trapped-ion quantum computer

We deploy a universal fault-tolerant quantum computing (FTQC) architecture based on error-detecting $[\![8,3,2]\!]$ and $[\![4,2,2]\!]$ codes on Quantinuum's Helios-1 trapped-ion quantum computer. We ...

arxiv.org

Me still in the honeymoon phase with my Chat.

Nicolas Addington@naddington.bsky.social · 3w ago

👀 arxiv.org/abs/2610.10072

OK, maybe we can't compute on galactic-sized matrices with infinite-precision complex numbers, but what if God wants to unitarily evolve the universe or compute its wave function in Fourier basis a bit more efficiently?

Ben Recht@beenwrekt.bsky.social · 2w ago

OpenAI’s faster Fourier transform is less a mathematical breakthrough and more a malicious exploit of mathematical traditions.

extremely concerning that OpenAI has made 3 sign errors—an odd number—upsetting this delicate balance

Daniel Litt@littmath.bsky.social · 4w ago

that the mathematical literature as a whole is so reliable comes down to the fact that, over time, we have made an even number of sign errors

Previously I used to print an interesting paper out and then carry it in my backpack for months without reading it. Now I can just ask an LLM to explain it to me and not read the explanation! Technological progress!!

Something to cheer you up! This is part of a campaign by the Norwegian Consumer Council against enshittification. www.youtube.com/watch?v=d6SP...

The Life of Death and the Ensh*ttificator

YouTube video by Forbrukerrådet - Norwegian Consumer Council

youtube.com

OpenAI: I proved that only 3 MUBs are possible in dimension 6. github.com/openai/math/... Me: Great, but where are the scripts your solution relies on? - seed.cpp - covout.cpp - ub.cpp - patchx.cpp - atlas.cpp - stageB.cpp - stageC.cpp - proof_run.py - ... OpenAI: Ups, forgot to attach...

dlyongemallo.bsky.social@dlyongemallo.bsky.social · 3w ago

Who here is an expert in MUBs and has looked at OpenAI's paper? Does it look legit? github.com/openai/math/...

Jokes aside, there's some serious discussion on this here. www.youtube.com/watch?v=k_Co...

Can AI Really Break Cryptography? Experts Weigh In

YouTube video by a16z crypto

youtube.com

Dulwich Quantum Computing@dulwichquantum.bsky.social · 3w ago

Cryptography used to rely on hardness assumptions. Now it relies on a "bad PR assumption": a cryptosystem is still safe to use since it would be bad PR if AI labs announced how to break it. Pro tip: Given the absence of cryptography problems in the OpenAI release, consider a career in cryptography!

Next time I can't sleep will try to read 168,000 lines of Lean code.

Maris Ozols@marisozols.bsky.social · 3w ago

You may finally rest easy at night since SICs exist in all dimensions! github.com/marozols/SIC... SICs are d² unit vectors in d complex dimensions with pairwise equal inner products. Their existences in all dimensions d was conjectured by Zauner in 1999. en.wikipedia.org/wiki/SIC-POVM

A website that lets you document how you got scooped by OpenAI.

Anuj Dawar अनुज दावर انُج داور@anujdawar.bsky.social · 3w ago

Hearing anecdotally of other such stories. Then there's this. curticapean.net/math-notes/

Cryptography used to rely on hardness assumptions. Now it relies on a "bad PR assumption": a cryptosystem is still safe to use since it would be bad PR if AI labs announced how to break it. Pro tip: Given the absence of cryptography problems in the OpenAI release, consider a career in cryptography!

No one is safe from making a sign error, even super intelligence! github.com/openai/math/...

If they now post 2 a month on @arXiv they are done in 60 years… 🙃

"Our strategy is not to decide which scientific problems deserve attention or try to solve them all ourselves. It is to put capable tools in the hands of the research community and let researchers pursue the questions they know best." --OpenAI, July 29, 2026 openai.com/index/chatgp...

OpenAI: "We’re releasing a broad range of new mathematical results produced by an internal frontier model." (722 manuscripts)

Me: Feeling empowered again. openai.com/index/sharin...

Dulwich Quantum Computing@dulwichquantum.bsky.social · 2mo ago

"A major goal of our work is to empower scientists to advance research and technology that benefits all of humanity." openai.com/index/navier... Then why does this make me feel so powerless?

Me: We're fucked, math is over! Also me: I can't even understand the statements of these results! Now we need more mathematicians than ever before! It's the golden age for math now!! github.com/openai/math

GitHub - openai/math

Contribute to openai/math development by creating an account on GitHub.

github.com

Me checking in on my agents from time to time.

`Da Elf@elfin.mstdn.social.ap.brid.gy · last mo.

Timeline cleanser of a parrot that found a traffic cam to ham it up for.

Successfully crossing the QIP finish line this year.

How to make money with your AI agents. (So you can pay for more tokens...)

FQxI@fqxi.org · 4w ago

In 2007, FQxI member Garrett Lisi offered $100 to anyone who could find three sign errors in his online physics notes. There were no takers until recently, when an AI agent emailed. He tells the story on QSpace News: qspace.fqxi.org/news/165333/...

"arXiv now limits submitters to up to two submissions per calendar month, with a limit of three total active submissions at any given time."

arXiv@arxiv.bsky.social · 4w ago

arXiv has updated our policy on rate limiting for all submitters. This update was made to fairly distribute moderator time & support the arXiv community of staff, volunteers, readers & authors. Please read our announcement to learn more: https://blog.arxiv.org/2026/10/01/updated-rate-limit-policy/

When everyone decides to post their paper on arXiv on the same day to ensure maximum visibility for their work.