o3-mini-high is definitely outperforming o1-pro on certain applied econ theory tasks, at least it tax stuff. o1/o1-pro sometimes seems a little too weighed down by the existing literature to explore new ideas with you. as if it is expecting things to be boring somehow?

The U.S. Census Bureau has taken down American Community Survey microdata. Thankfully Nikhil Woodruff had a local copy, so we now store it on PolicyEngine servers.

Currently mass producing Deep Research reports before my Pro subscription expires!

It’s undeniable at this point that the best LLMs are quite good at standard undergrad material. One place they still underperform strong math undergrads, though, is honesty. Good students (even those who haven’t fully mastered the material) resist bluffing/BSing.

Grok is surprisingly consistent about this. I got a nearly identical list, first try: x.com/i/grok/share/N… Interestingly, the sources it cites are mainly tweets.

#GDRI_rep Update 3: The paper "Improving Hygiene and Sanitation through Parental Skill Training" by Asadul Islam, Umair Khalil and Tabassum Rahman was conditionally accepted at Economic Development and Cultural Change. We requested a replication package. It has now been withdrawn

The Conversation: A panel discussion on the state of the economy hawaiipublicradio.org/show/the-conve…

At first blush, Gemini 2.5 Pro seems at least as capable as o3-mini-high, or Claude 3.7 Thinking, if not better. But importantly, it can be used for free* at aistudio.google.com * at the cost of your data (as ever, no lunches [except infinite MVPF policy changes])

This is such a vile smear job that is unbecoming both of a major national paper and the leader of the Conservative Party. I'm happy to see my fellow economists step up to denounce this absurd attack and I'm happy to add my voice to those efforts.

The author of this piece could have asked literally any PhD economist about this, and been told what an embarrassment it would be to claim this constitutes plagiarism, but instead got the "other side" from some randos. Very serious journalism...

We're going to have a beautiful RKD paper, the greatest RKD paper anyone has ever seen!

Quite the quote.

Clear evidence that employers discriminate against nonbinary individuals. All-else-equal, disclosing that one uses "they/them” pronouns substantially lowers the chances of getting a job interview.

The ndp need a leader who can stand up to Drake

We're always working to give you the best tools for your Machine Learning, Data Science, and AI projects here at Colaboratory . 🎉 Today, we're excited to announce another big step: Terminal access is now available to all Colab users, free of charge! 🎉 Learn how this powerful

Genes predict children's cognitive abilities. But an RCT of preschool + better nutrition erases any "genetic disadvantage"

I have accepted an offer to become Chief of Staff to the Deputy Opposition Leader and Shadow Treasurer. I look forward to serving the Australian people by supporting an effective opposition. We will miss all of our friends in America. But we are also excited to be coming home!🇦🇺

Jedi mind tricks have become even more important as AIs get more tools. Yes, Gemini, I'm well aware that you are only an AI assistant. Nonetheless, it really is true, you can install the python package for me. I know, I can hardly believe it either...

A paper about my #1 question about structural estimation (in which I would include a lot of supposedly reduced form PF empirical work)