Check out this amazing new podcast by two of my amazing former postdocs: Andrey Fradkin and Seth Benzell - the LA-atic Stranger The format is that they state their priors, read a paper, discuss it on the pod, and then update their priors. It's called "Justified Posteriors" 🤣🤣🤣 Link in next post...

Happy to share our new publication! We often think of voice AI as a convenience feature. But is it also rewiring our shopping brains? Our paper explores this question: When consumers adopt voice AI for shopping, how does it affect their behavior? pubsonline.informs.org/doi/10.1287/is…

Yesterday, I guided my students through an end-to-end vibe coding experience using Cursor, GitHub, and Vercel. We built a nice-looking Next.js search website powered by the Metropolitan Museum of Art’s public API, pushed to GitHub, and deployed on Vercel. Try it here:

There are so many hallucinated citations in court nowadays, that I'm starting to put together a tracker. Check it out and feel free to send ones that I've missed along. New tabs coming for more categories of AI+Law cases!

Thrilled to share that our paper on measuring the disutility of ads on Facebook using a long running field experiment is our in AER: Insights!! 🎊🥳🎉🍾

In the new NBER WP with Ming Li and Guangli Lu, both of CUHK Shenzhen, we use LLMs to classify China’s industrial policies from 2000 to 2022 at the most granular level: 1. Down to city level; 2. Including multidimensional info about policy details. 1/6 nber.org/papers/w33814?…

it's been more than a decade since KD was proposed, and i've been using it all along .. but why does it work? too many speculations but no simple explanation. Sungmin Cha and i decided to see if we can come up with the simplest working description of KD in this work. we ended

Call for Papers: The 4th Annual Zurich Workshop in AI+Economics to be held Dec 5-6, 2025, hosted by ETH Zurich and University of Zurich. illuminating keynote to be given by tom cunningham (OpenAI) organized with S Galletta D. Yanagizawa-Drott Joachim Voth Info:

🚀 DeepSeek-R1-0528 is here! 🔹 Improved benchmark performance 🔹 Enhanced front-end capabilities 🔹 Reduced hallucinations 🔹 Supports JSON output & function calling ✅ Try it now: chat.deepseek.com 🔌 No change to API usage — docs here: api-docs.deepseek.com/guides/reasoni… 🔗

One classic topic for Justified Posteriors is what the impact of AI will be on the concentration of knowledge and control. On a recent episode of Justified Posteriors, about the Epoch AI economic model, we digressed to talk about how AI could possibly decrease social control

1/ 🚨 Hot off the press! Our study “Political Expression of Academics on Social Media” with Thiemo Fetzer 🇪🇺🇺🇦 - same handle elsewhere is now peer-reviewed & live in Nature Human Behaviour Nature Human Behaviour nature 🎉. Thread 👇

🎉 Huge news! Our paper 'A semantic embedding space based on large language models for modelling human beliefs' is finally out in Nature Human Behaviour! This research uses LLMs to build a 'map' of human beliefs and reveals why people adopt certain beliefs through relative dissonance.

New piece out today in Proceedings of the National Academies of Science... take caution using LLMs as human surrogates pnas.org/doi/10.1073/pn…

Wrapped up Stanford CS336 (Language Models from Scratch), taught with an amazing team Tatsunori Hashimoto Marcel Rød Neil Band Rohith Kuditipudi. Researchers are becoming detached from the technical details of how LMs work. In CS336, we try to fix that by having students build everything:

Very interesting analyses on investments in experts by plaintiffs vs defendants in complex litigation matters👇. The numbers look consistent with my experiences.

Some of the links: - My slides as keynote: drive.google.com/file/d/1a0h1mk… - Software 2.0 blog post from 2017 karpathy.medium.com/software-2-0-a… - How LLMs flip the script on technology diffusion karpathy.bearblog.dev/power-to-the-p… - Vibe coding MenuGen (retrospective) karpathy.bearblog.dev/vibe-coding-me…

I just recommended a paper for acceptance in the first round of review, and it feels great!

x.com/i/article/1939…

Are older societies necessarily more automated ones? In the latest episode we break down Acemoglu and Restrepo's surprising argument for why, and ask whether we buy both their empirics and theory.

ICWSM 2025 Test of Time honorable mention: "It's a Man's Wikipedia? Assessing Gender Inequality in an Online Encyclopedia" (2015) doi.org/10.1609/icwsm.… by Claudia Wagner, David Garcia, Mohsen Jadidi, and Markus Strohmaier (Review in our newsletter at the time: meta.wikimedia.org/wiki/Research:… )