
Ever wondered why presenting more facts can sometimes *worsen* disagreements, even among rational people? 🤔 It turns out, Bayesian reasoning has some surprising answers - no cognitive biases needed! Let's explore this fascinating paradox quickly ☺️
✓Past: 👨🔬 RS DeepMind 2y, Midjourney 1y 🧑🎓 DPhil AIMS 4.5y 🧙♂️ RE DeepMind 1y 📺 SWE Google 3y 🎓 TUM 👤 @nwspk

Ever wondered why presenting more facts can sometimes *worsen* disagreements, even among rational people? 🤔 It turns out, Bayesian reasoning has some surprising answers - no cognitive biases needed! Let's explore this fascinating paradox quickly ☺️

We’re releasing all 57 intermediate checkpoints from Marin “Mochi” 535B-A23B pretraining, including 10 from long-context extension. 18.2T tokens, about 61 TB of weights. huggingface.co/open-athena/M… Blog: openathena.ai/blog/marin-moc…

In my ongoing quest to find the limits of Opus 5.5, I directed it to audit my old scientific field, theoretical astrophysics. We indexed about 150,000 papers, audited 50,000 for issues, and did progressively more thorough checks on thousands more. We found nearly 1000 major errors that got cited many times, most are still uncorrected. This survey enabled me to estimate the efficacy of peer review for correcting errors (not very good) and to identify dozens of open questions that anyone with spare Claude Max tokens could answer today (.md context files included). Bottom line: AI can speed up quantitative science by >1000x with no further model improvement required. But I'm seeing no evidence of this being pushed within universities, only academics complaining about labs dropping math proofs. Among my fellow reformed academic high agency founders, however, several are speed running entire disciplines. This survey has implications for how science will be run in the very near future. Hint: Academic publishing is dead. It was already a very stupid way to record and promulgate knowledge, and had been so for decades, even before all that publicly funded knowledge was locked behind paywall cartels. Now AIs can generate thousands of novel, correct papers per day, it is no longer meaningful for humans to consider a research paper a "unit of science" or to even bother much with the 99.9% of science which, up to this point, was necessary drudgery. Time to discover something like General Relativity or Evolution every week. LFG!! Enjoy: science.caseyhandmer.com/phy…

The question is not whether it's just repeatable but also if it is reproducible. That is good science

Introducing Qwen3.8-27B-pi-adapted 🤗 We adapted Qwen's model specifically for the pi coding agent using reinforcement learning and a LoRA adapter. The results, measured as pass@1: 🚀 SWE-bench Pro (Hard 51): 29.4% → 62.7% 💻 Terminal-Bench 2.1: 39.5% → 55.8% ⚡ 55–62% fewer tool calls per task This is the power of open models: you can adapt them to YOUR tools and YOUR use cases. Let’s build! 💛 Open weights: huggingface.co/harness-hub/Q… Methodology: huggingface.co/blog/harness-…

NEW ⚡: We let Opus 5.5 run an AI company for seven weeks. We gave it an HF account (Horizon-Labs) and compute, and asked it to earn real adoption by creating useful models. It reached 50k downloads and its models are being used in more than 100 GitHub projects.

Excellent evaluations as always coming from @AISecurityInst 👏👏👏

Going to be difficult given pre-training is illegal in the UK.

With Carbon's help, I ran Gemma directly in an office printer 😅🖨️ 🤏Carbon compiled a small transformer engine + quantized Gemma 3 270m into a single PostScript job ⚡️ 100% in the printer: 14MB peak RAM and blazing 60s per token 🧠Smart layers juggling to avoid OOMing

The week of my departure from OpenAI, I saw external auditors negotiating with OpenAI over small improvements to access and scope to the company's safety-relevant information, and asked my colleagues “Why are we having these conversations at all?” — @sama already publicly committed to continued, employee-like access for external assessors — and these negotiations sounded not even close. Nobody had a good answer. Today, OpenAI announces the engagements they’re doing with METR, Apollo Research and others — and as many have already pointed out today, it looks like OpenAI employees were not able (although I'm sure they tried) to hold the leadership to @sama’s public commitments: * No Slack access (for context, all OpenAI safety discussions and debate happen over Slack; anything that misses these conversations will lack important insights) * OpenAI picks employees that assessors are allowed to talk to. * If an employee raises a concern to assessors, they cannot disclose it publicly even if it doesn’t contain confidential information, unless assessors themselves verify it or OpenAI volunteers their own statement confirming the employee’s concern. Assessors cannot quote employees, even anonymously. * Some of the external assessors (not sure if all?) are barred from publishing any results about internal-only models (this means no new results for the model responsible for the Hugging Face incident; and if internal-only models are used for RSI, no alignment assessments of the world's most dangerous models) Beyond the object-level details, if you read posts by each of the third party groups, they all say their terms do not meet the scope and access that they themselves proposed previously. Unless OpenAI has concrete plans for reaching the employee-like bar in the immediate future and just forgot to say that in the post, this looks very unfortunate. To those who hope we'll get there someday, I remind that time is short: last week some of the best AI forecasters updated their superhuman researcher timelines from a range between 2028-2033 to mid-end next year.

Crazy...this is Kubernetes moment for AI Agents. Run millions of agents on Kubernetes. Idle agents go to sleep and wake up in under 500ms. 10x more agents on the same hardware. 100% Opensource. github.com/agent-substrate/s…

DocuSign wants $8,100/yr from a 15-person team just to sign PDFs. So I built SignuDoc: open source, one-click deploy to your own @vercel account, $0 per seat. White-label branding, full audit trails, tamper-evident seals and your documents never leave your cloud.

Sounds like they are employee-like after all

Really excited to share these opportunities to join @CompSciOxford and work with us to shape the future of open AI! We have 3 senior postdoc positions, and 2 research lead positions to build a group under you Deadline is 27/11 my.corehr.com/pls/uoxrecruit… my.corehr.com/pls/uoxrecruit…

The oddest intellectual position I have encountered over the last decade is the refusal of so many to believe that the current version of Pangram is incredibly good at avoiding false positives (i.e., flagging a human-written text of some length as AI-generated). It is not terribly hard to trick Pangram into false negatives (i.e., classifying an AI-generated text of some length as human-written) with “humanizing” tools and some careful editing. But the other way around? Nearly impossible. For two weeks, I have been running experiments in many different situations for a new paper I will circulate tomorrow (and because, as an editor, I care a lot about the answer). Not in a single case did a human-written text of reasonable length score above 10% for AI generation. And, yes, many the texts I used (including mine!) are from ESL writers. Frankly, the counterarguments I have read over the last week here on X (and the papers linked to “back them up”) are childish in the extreme. If you think that false positives at Pangram are, as of November 16, 2026, a concern, you are lying to yourself.

Really cool to see more results like that This makes sense intuitively. Models make mistakes when generating autoregressively and with noising they learn to auto-correct during pretraining Hypothesis: their own generations become less OOD (RL fixes a similar issue as well) (Also connections to information bottlenecks etc)