"We are investing so much in driving the capabilities for hundreds of billions of dollars, and we're investing very little in ensuring that leads to widely shared prosperity." Director Erik Brynjolfsson discussed AI and future of work with NPR's Steve Inskeep. npr.org/2025/08/05/nx-…

left: violent crime in the usa and uk (tanking) right: the share who believe crime is on the rise (steady/growing)

Really incredible that Daniel Rock and OpenAI's research in 2023 forecasted which jobs would eventually use GenAI. This is strong support for the task based approach to work used by David Autor Erik Brynjolfsson and others. Jobs might just be bundles of tasks!

Using AI doesn’t have to make us dumber. Evidence: AI boosts service rep productivity and customer satisfaction. Even during AI outages, those who use it regularly are more effective. Outsourcing thinking to AI yields cognitive debt. Learning from AI pays cognitive dividends.

My longtime collaborator Dave Patterson (long-time faculty at UC Berkeley, Association for Computing Machinery Turing Award winner, and fellow Laude Institute board member) wrote a very good op-ed about how continued investing in basic science and technology research is essential for the U.S. Dave

Argentina went completely off track because of decisions made in a few years by an elected — through a democracy — leader who pursued autocracy rather than venerating democracy. And that should be a cautionary tale for everyone in the business community and everyone involved in

Just read the new Dan Wang book Breakneck. It is an extended version of his annual letters from China: original reporting, details I don’t get anywhere else, and smart insights that will be repeated in MSM later. So I’m going to host a chat with Dan and Noah Smith 🐇 a few days

After years of inaction, we’re finally taking action on lowering housing costs. There’s more to do, and I’m ready to keep leading the charge.

Lab Director Erik Brynjolfsson spoke with The New York Times's Larry Sommer about tracking labor statistics -- read his thoughts at the link! nytimes.com/2025/08/08/bus…

I'm noticing that due to (I think?) a lot of benchmarkmaxxing on long horizon tasks, LLMs are becoming a little too agentic by default, a little beyond my average use case. For example in coding, the models now tend to reason for a fairly long time, they have an inclination to

Never leaving this website, roon

A 457-page Stanford 2025 AI Index Report - Costs fell fast, GPT-3.5-level inference dropped 280x, hardware -30%/year, efficiency +40%/year, open-weight gap 1.7%. - Training compute doubles every 5 months. - Optimism rose but divides are large, China 83% vs US 39% see net

never let the members of technical staff name things

Goldman's latest (still very early) analysis of tariff effects thru June 2025: -Foreign exporters absorbed 14% of US tariffs -US companies ate 64% -US consumers ate 22% -Protected US companies also raised prices -Consumers will see bigger price increases (70%) thru the Fall

Testimonials are coming in from teachers about their newly phone-free schools, and the amazing things that students are now doing. Like taking notes, completing their assignments, then talking to each other. "Was it this easy of a solution the whole time?"

What's the best explanation for the 20-30 year lows in crime rates in Washington, DC, San Francisco, NYC and other major US cities?

"Simply strolling around New York makes nativism intellectually and emotionally hard to sustain.  Ubiquitous foreigners educate far more effectively than holier-than-thou preaching ever could." betonit.ai/p/natives_are_…

"What country has ever suffered from cosmopolitan tolerance run amok?  From focusing on people’s common humanity rather than superficial differences?  From judging people on their merits instead of their origins?" betonit.ai/p/natives_are_…