How to extract theoretical insights from machine learning algorithms, from 𝐒𝐞𝐧𝐝𝐡𝐢𝐥 𝐌𝐮𝐥𝐥𝐚𝐢𝐧𝐚𝐭𝐡𝐚𝐧 and Ashesh Rambachan nber.org/papers/w32422

🌎 Are you ready to solve one of the world's biggest challenges? We're excited to announce a new round of #100andChange – a competition for a single $100M grant! Proposals must identify a problem & offer a promising solution for change. Learn More ⬇️ macfound.org/programs/award…

Increasing numbers of people claim that current AI systems "understand" and "reason" through the problems they are given. Or at least are on the cusp of it, so that Artificial General Intelligence is right around the bend. I don't think that's the case. Here is why.

Really excited to see our paper w/pamela mishkin @thankyourniceAi sam manning out today in @sciencemagazine. Link here: science.org/doi/10.1126/sc… tl;dr: Studying where LLMs might affect workers, we find that GPTs are GPTs! But what does that mean and how did we do it? 🧵

Excited to share something I was working on for a while now: the Before AGI podcast! There is a lot of talking about AGI. But what’s less explored is an equally important question: What happens before AGI arrives? And what should we be doing to prepare? First episode with

Behavioral economists have a lot to offer in the building and evaluation of LLMs. Exhibit 1:

LLMs don’t behave like people, even though we may expect them to. A new study from LIDS PIs Ashesh Rambachan and 𝐒𝐞𝐧𝐝𝐡𝐢𝐥 𝐌𝐮𝐥𝐥𝐚𝐢𝐧𝐚𝐭𝐡𝐚𝐧 shows someone’s beliefs about an LLM play a significant role in the model’s performance and are important for how it is deployed. bit.ly/3y2zmRm

Claudia Goldin is not only an amazing scholar but also a tremendous southpaw pitcher as seen in her perfect strike first pitch for the Red Sox-Yankees game on 7/26

Lists of biases point out problems. Developing a "pattern language" could point us to solutions, writes ideas42's Piyush Tantia. Drawing inspiration from architecture and UX design, he asks: "Can we create a pattern language for behavioral design?" behavioralscientist.org/can-we-create-…

Cool workshop at behavioral econ/ML boundary:

Great job market paper on topic close to my heart: algorithmic hypothesis generation. (As a bonus, I get be proud to have had the chance to advise a student as creative as Raf)

Cool write-up by Joshua Gans joshuagans.substack.com/p/knowing-when…

This was such a fun conversation to have - even if it is 50% as fun to listen to, it's worth the listen! (also Aleksander Madry - great podcast title!)

Our series highlighting Steve Levitt’s ten favorite episodes continues: We revisit Steve’s 2021 conversation with economist Sendhil Mullainathan (𝐒𝐞𝐧𝐝𝐡𝐢𝐥 𝐌𝐮𝐥𝐥𝐚𝐢𝐧𝐚𝐭𝐡𝐚𝐧) about how to make memories stickier, and how to find something new to say about scarcity. freak.ws/4cWLDoR

🎙️Just finished listening to the latest episode of Before AGI from Aleksander Madry with the brilliant 𝐒𝐞𝐧𝐝𝐡𝐢𝐥 𝐌𝐮𝐥𝐥𝐚𝐢𝐧𝐚𝐭𝐡𝐚𝐧 . If you're interested in algorithmic bias and fairness, this episode is a must💡 A few selected quotes 👇

Excited for the official announcement of HAI Lab at the University of Oxford, due to launch in the Fall, which I will direct as the inaugural McCord Professor of Philosophy and AI. 🧵

Incredibly important piece by Jens Ludwig in the Chicago Tribune. "The most important problem in higher education that no one is talking about," including this devastating chart: chicagotribune.com/2024/09/08/opi…