We’re livestreaming the Joint California Summit on Generative AI on Wednesday, May 29. Join us alongside Berkeley Computing, Data Science, and Society, GO-Biz, and CA Gov Ops to discuss the impacts of GenAI on California and its workforce. #CaliforniaGenAI

Watch here at 9 am PT: stanford.io/3KjzpKX

LLMs could redefine the future of healthcare delivery, but so far there’s a big gap between promise and practice. Stanford scholars looked at the latest studies and concluded we need more rigorous evaluation. stanford.io/3wPSily

Is AI sentient? My friend and colleague Prof. John Etchemendy, a renowned professor and co-Director of Stanford HAI , just co-authored this piece to debunk the claim that today’s LLMs are sentient TIME
time.com/collection/tim…

“It is important to understand the profound difference between how humans generate sequences of words and how an LLM generates those same sequences.” Stanford HAI co-founders Fei-Fei Li and John Etchemendy debunk the idea that LLMs have achieved sentience. bit.ly/4bP44LV

Can you guess the 6 biggest players in AI? See the list here, along with a dozen other highlights from the #AIIndex2024 Report. stanford.io/4dPLODZ

Application is now open for the Stanford HAI Congressional Boot Camp on AI – now in its 3rd year! The bipartisan program equips congressional staff with comprehensive knowledge needed to think critically about regulating the technology. Apply by June 7: stanford.io/4bNQEjh

Das HPI und das Stanford HAI wollen in einem gemeinsamen Forschungsprogramm wissenschaftliche Projekte zu menschenzentrierten #KI -Systemen entwickeln. Für einen Kick-off Workshop waren Forschende aus Stanford in dieser Woche am HPI zu Gast. 💡Mehr Infos: hpi.de/news/jahrgaeng…

“To ensure that models are safe for widespread adoption, we need to develop a science of AI safety,” HAI senior fellow Rob Reich told in a recent interview. “This is one of the most important missions of the AI Safety institute.” (via Stanford Impact Labs) stanford.io/4btIUTO

On @standordeng’s The Future of Everything podcast, cybersecurity expert and Stanford HAI associate director Amy Zegart says foreign access to American data is a national security threat. stanford.io/4aMfMGl

As more nations consider integrating autonomous AI agents in high-stakes military and diplomatic decision-making, a new Stanford HAI policy brief explores the risks. stanford.io/4bPi5sO

Today we released Foundation Models Transparency Index (FMTI) v1.1, 6 months after v1.0, which tracks companies’ level of transparency.
x.com/RishiBommasani…
I want to draw attention to the companies’ transparency reports that FMTI v1.1 is based on:

Two Stanford University AI projects from Stanford UniversityEng and Stanford UniversityMed were among the 35 research teams selected to receive grants for access to advance computing resources through the #NAIRR pilot. Learn more about their proposals: Dorsa Sadigh Kevin Schulman stanford.io/3wH67Tn

Explore the foundations of modern deep learning methods for learning across tasks in our upcoming online course taught by Chelsea Finn. Course starts June 3 and runs 10 weeks. Enroll ➡️online.stanford.edu/courses/xcs330… Stanford AI Lab Stanford HAI #AI #ML

Following an earlier assessment six months ago, a new Foundation Model Transparency Index by Center for Research on Foundation Models indicates a notable shift: developers are now disclosing more information about their practices. crfm.stanford.edu/2024/05/21/fmt…

Foundation model developers are disclosing more information but there is plenty more to do.

We are releasing the May 2024 Foundation Model Transparency Index where we score 14 prominent developers on our 100 point Index.

crfm.stanford.edu/fmti/

Researchers are turning to LLMs for help with writing academic papers. In a recent study, Stanford University scholars investigated how often it’s happening. stanford.io/3QTYlwk

AI propaganda is here. But is it persuasive?

Dylan Walsh wrote about Josh Goldstein's co-authored paper on AI-generated propaganda for Stanford HAI.
hai.stanford.edu/news/disinform…

It’s difficult to build trust in AI medical programs due to their black box nature. To better understand their internal reasoning, researchers leveraged human expertise alongside another kind of artificial intelligence: generative AI. stanford.io/3QR6nWy

At the inaugural #RAISEHealth symposium, speakers explored how to integrate AI into healthcare in a way that’s beneficial to both physicians and patients. Responsible AI requires collaboration among multiple stakeholders, said Fei-Fei Li. (via Stanford Medicine) stan.md/4bHduJd

Just how persuasive is AI-generated propaganda, and how can we mitigate the potential dangers? Stanford HAI faculty affiliate Michael Tomz and colleagues explore these questions in a new study. stanford.io/44TIxPR