Google has integrated Google Scholar with AI. Run a search and it will screen the 10 most relevant papers and shows how the paper answer your question. This will make literature discovery very efficient. Here's how to use it:

Thanks for the great conversation Matt!!

Im grateful to all of ML community. The stuff we are collectively through is mindbending and it only gets weirder from here. It’s much better to face difficulties of safe deployment of machine intelligence together and there are tons of genuinely good people in the world who

🚨 AutoBench 1.0 – Run 4 is LIVE 📷 - 33 frontier models ranked (including GPT-5.1, Gemini 3 Pro, Grok 4.1, Kimi K2 Thinking, etc.) - 21 ranking models - 300+ fresh questions generated - 220,000+ individual rankings This is the most manipulation-resistant evaluation we’ve ever

Social media tends to frame AI debate into two caricatures: (A) Skeptics who think LLMs are doomed and AI is a bunch of hype. (B) Fanatics who think we have all the ingredients and superintelligence is imminent. But if you read what leading researchers actually say (beyond the

We are on the cusp of a profound change in the field of mathematics. Vibe proving is here. Aristotle from Harmonic just proved Erdos Problem #124 in Lean, all by itself. This problem has been open for nearly 30 years since conjectured in the paper “Complete sequences

ChatGPT is 3 years old. This was the TikZ's unicorn on the day it came out. Pretty wild progress in 3 years.

I'm at neurips this week c: I don't have any papers to present. But in the past 3 years, I had: been part of openai's reasoning effort, designed the initial (dumb) versions of what is now our (much more fancy) internal reasoning algorithm, worked on the o1/o3 scale-up, and am

Thrilled to share that Anna Goldie and I are launching Ricursive Intelligence, a frontier lab enabling recursive self-improvement through AIs that design their own chips. Our vision for transforming chip design began with AlphaChip, an AI for layout optimization used to design four

We trained a variant of GPT-5 Thinking to produce two outputs: (1) the main answer you see. (2) a confession focused only on honesty about compliance. The main answer is judged across many dimensions—like correctness, helpfulness, safety, style. The confession is judged and

8th of December 1903 the new York times published this. 9 days later we had the first flight. Don't say AI won't do, you will probably be wrong.

Separate reports by the publicity firm Edelman and Pew Research show that Americans, and more broadly large parts of Europe and the western world, do not trust AI and are not excited about it. (Links in original text, below.) Despite the AI community’s optimism about the

GPT-5.1-Codex Max is now available in the Responses API. First released in Codex two weeks ago, our most capable agentic coding model is now available to integrate into your apps and workflows. If you use the Codex CLI via API key, you can now also use GPT-5.1-Codex-Max.

The new Codex model is available in Cursor! It's free to use until December 11th. We worked with OpenAI to optimize Cursor's agent harness for the model. cursor.com/blog/codex-mod…

🚨 New in Nature+Science!🚨 AI chatbots can shift voter attitudes on candidates & policies, often by 10+pp 🔹Exps in US Canada Poland & UK 🔹More “facts”→more persuasion (not psych tricks) 🔹Increasing persuasiveness reduces "fact" accuracy 🔹Right-leaning bots=more inaccurate

Pope Leo is right: "How can we ensure that the development of artificial intelligence truly serves the common good, and is not just used to accumulate wealth and power in the hands of a few?” We must demand that the benefits of this technology work for all, not just the rich.

I complain a lot about RL lately, and here we go again. The CS view of RL is wrong in how it thinks about rewards, already at the setup level. Briefly, the reward computation should be part of the agent, not part of the environment. More at length here: gist.github.com/yoavg/3eb3e722…

I'm seeing lots of confusion about ads rumors in ChatGPT. There are no live tests for ads – any screenshots you’ve seen are either not real or not ads. If we do pursue ads, we’ll take a thoughtful approach. People trust ChatGPT and anything we do will be designed to respect that.

As an AI expert who totally doesn't think LLMs are the path to AGI, but uses LLMs very avidly to get things done, here is my take on the utility of top LLMs currently: 1) If you don't know what you're doing in a certain domain, you can get interesting and valuable results, but

Around 12 years ago I came to the Bay to work for Google and heard about the dream of self-driving cars. I already spent way too much time in Waymo in San Francisco but only today got off the highway waitlist. Finally it can drive around the Bay, what an incredible achievement!!