Should AI models be “open” or “closed”? Today Carnegie joins with leading voices from across the spectrum to help advance this historically fraught debate. Our paper tries to identify emerging common ground and to highlight key unanswered questions: carnegieendowment.org/research/2024/…

4. Marginal risk is the right framework but we need much better pre- *and* post-release assessments. 5. No clear evidence yet that major AI companies have erred in openly releasing any foundation model, but we can’t say for sure.

Tune in to CNN @ 10am ET today for a GPS special, “Reengineering Life: The Next Frontiers in Science” I’ll examine two technologies—gene editing & artificial intelligence—in interviews w/ two women who pioneered them: UC Berkeley’s Jennifer Doudna & Stanford’s Fei-Fei Li

.Fareed Zakaria has a terrific show this morning featuring two leading women in technology and science: Jennifer Doudna Doudna Lab (from U.S. Department of Energy’s LBNL) and Fei-Fei Li Fei-Fei Li (from Stanford University) — solving the most vexing problems in health and AI. Inspiring!

General Purpose Technologies, like AI, have effects that are complex; impossible to predict in advance; impact different groups & industries differently; and which play out over time. The sooner we get past AI is “all good” or “all bad” arguments & focus on context, the better.

“The question of whether open foundation models are dangerous has been a little bit oversimplified and kind of alarmist” (@Random_Walker, Princeton Center for Information Technology Policy). ➝ Spotify: spoti.fi/4g2Cf5y ➝ Apple: apple.co/3AHx4rF ➝ YouTube: youtu.be/stJaSvQoIVQ #scalingtheory

Rather odd commentary in the Financial Times today. The main case in point seems to be that the compute power threshold for designating GPAI models with systemic risk might soon be outdated. Still, it ignores that the Commission can revise it via a delegated act. 1/2 ft.com/content/773eb1…

HELM goes multilingual! We collaborated with Liwei Wang 's team to get CLEVA (arxiv.org/pdf/2308.04813), a Chinese LM benchmark, into the HELM framework. crfm.stanford.edu/helm/cleva/

Designing thresholds for policy is hard. Here are my thoughts: crfm.stanford.edu/2023/11/18/tie… I agree with Luca Bertuzzi that this article makes an absolute farce out of serious policymaking by the EU legislators on the AI Act. Let alone Annex XIII + Scientific Panel qualified alerts...

This is is exactly the kind of rhetoric I was speaking out against earlier.. The debate around SB1047 is not Dan Hendrycks v Big Tech. There is a legitimate diverse coalition of perspectives on either side! And opposing the bill does *not* default to an anti-regulatory stance.

Thank you so much TIME magazine for recognising research on the ethics of AI—undertaken with colleagues—in this year’s Time AI100 list! 🎉 #TIME100AI time.com/7012861/iason-…

Ben Recht His position on open source AI is *very* different from his open internet era.

What an amazing 10 years with Caltech & Yue Lab Caltech! I'm very grateful for the wonderful students, postdocs, and other collaborators I've worked with. Thanks to everyone who attended, especially the alumni who traveled back to celebrate! Listening to the stories that were

#CornellTech warmly welcomes Assistant Professor of #CS Hadar Averbuch-Elor, Clinical Professor of #Law and Research Director David Reiss, and Assistant Professor of #IS Angelina Wang as new #faculty! bit.ly/3ZfC4xQ #EducatingLeaders Cornell University Hadar Averbuch-Elor Angelina Wang

Years ago people working in AI policy (including me) perhaps foolishly wanted to 'wake up' DC. Well, we got what we wanted! Now I think a lot of the challenge of AI policy relates to trying to stabilize this leviathan that has been woken.

This is very nice to see and richly deserved! Iason has been a quiet leader in this space for a long time, and besides his own excellent work has also brought a lot of younger folks through. Many other great people on this list as well—impressed with the research behind it.

Many people have been saying things like this, but this is quite false. 500m in damage isn't the end of the world -- it's a Tuesday in the global economy. Let me give examples.

We had the incredible opportunity to sit down with Sarah Bana , Assistant Professor of Management Science at Chapman University and Digital Fellow at the Stanford University Digital Economy Lab, on the latest DataBytes podcast! In this episode, we explore: 🔍 How new

Large Language Model Agents is the next frontier. Really excited to announce our Berkeley course on LLM Agents, also available for anyone to join as a MOOC, starting Sep 9 (Mon) 3pm PT! 📢 Sign up & join us: llmagents-learning.org

Very fun conversation with Sadie, talking about what we are seeing so far with respect to AI and the labor market, and what some very smart folks (Erik Brynjolfsson, Matt Beane, and Daniel Rock) are telling us to look out for!!