yea, we’re cooked.

How I’ve been feeling about the current state of ML research: bosonphoton.github.io/thoughts/itsdo… TLDR; It has converged to the same few topics within NLP. And it’s boring.

We are bringing back Stanford’s CS 25 Transformers Course (cs25.stanford.edu) today! It’s open to everybody! This is one of Stanford University's hottest seminar courses. We open the course through Zoom to the public. Lectures start today (Tuesdays), 3-4:20pm PDT, at

Stanford CS25 Talk Today: Eshed Margalit, Noetik.AI Today (Tues, 5/20) at 3pm PDT, Eshed Margalit from Noetik.AI will be giving a talk for CS25 (cs25.stanford.edu). The talk will be livestreamed on Zoom at stanford.zoom.us/j/91661468474?…. Presentation

Fully packed first class! Excited to TA Stanford CS329A, Self-Improving AI Agents, with Aakanksha Chowdhery Azalia Mirhoseini Chelsea Zou. Here's to an amazing quarter!

Very excited to teach CS329A: Self-Improving AI Agents, with Aakanksha Chowdhery for the second time in 2025! Course website: cs329a.stanford.edu

Vibe coded an AI poker coach. Track the hands you play during live sessions and get feedback from AI. gbowizard.vercel.app I am not liable for any losses, questionable poker advice, or sudden gambling addictions that arise from the use of this app.

me at house parties

Everyone loves Claude Opus. Poker did not. Introducing Poker Agent ❤️♠️, a new LLM poker benchmark built with Vals AI. We pit 17 frontier LLM agents against each other across over 20,000+ hands of poker. Claude 4.5 Opus placed in the bottom half while GPT 5.2 and Gemini 3

[CL] A Unified Definition of Hallucination, Or: It's the World Model, Stupid E Liu, V Gangal, C Zou, X Huang... [CMU & Patronus AI & Stanford University] (2025) arxiv.org/abs/2512.21577

Dan Roy We have the v1 of a (hopefully) grand unified defn. by requiring added specification of a reference world model here: arxiv.org/abs/2512.21577 Glad to hear thoughts or feedback!

every time i try to have an original research idea but Zhang et al. already published it 3 years ago

Are we prompting the agents or are the agents prompting us…

anything good that generates code specifically from papers yet

what a time to be alive

x.com/i/article/2032…