More excited than ever to announce $1.7M:
'CBS-NTT Program in Physics of Intelligence at Harvard'! 🧠

With new technology comes new science. The time is ripe to build a better future with 'Physics of Intelligence for Trustworthy and Green AI'!

🧵👇news.harvard.edu/gazette/story/…

Last week at the AE Spring Summit we hosted a fascinating panel on Fundamental Technical Challenges for ML/AI:

youtu.be/XCUq9Yq1BS8?si…

Many thanks to panelists:
Gayatri Sarkar (Gayatri Sarkar ✊🏽🌏, Advaita Capital 🌏🌲)
Eliza Kosoy (@ElizaKosoy, Berkeley AI Research)
Surya Ganguli (@SuryaGanguli,

An excellent thread on homoclinic chaos and more!

Looking forward to speaking to the National AI Advisory Committee (ai.gov/naiac/) tomorrow April 16 12:30pm PST on advancing biology and human health with AI, and related regulatory issues.
To attend virtually:
events.nist.gov/profile/form/i…
Agenda:
nist.gov/system/files/d…

Reviewers / meta reviewers: this is an example of what NOT to do. Isolating the simplest possible failure modes of complex models is a worthwhile contribution - as is (in other cases) deriving the simplest possible model of a phenomenon that makes testable predictions.

Here is a meta-review we got for third submission of a paper that aims to study text-understanding capacities of LLMs, focusing on very simple, if not trivial, cases where they systematically fail.

We see every stated weakness as a strength, and they are all by design.

What are some of the Grand Challenges in Healthcare that AI gives us a shot at addressing? Come along with Vijay Pande and I as we jam on:

➡️The claim as a bundle of logic to be automated.
➡️Always-on clinical trial A/B tests at population scale.
➡️Spot pricing for healthcare

I wrote a short 'perspective' piece about some of the challenges we face in building generative models for physical and chemical systems. The main contention is that incorporating physics more directly will help! authors.elsevier.com/a/1itv4_UvfPrb…

Should be fun - come join this summit on fundamental challenges in AI!

'Beauty is truth, truth beauty—that is all Ye know on earth, and all ye need to know.” So it goes for art, life, science, math, and yes… well executed startups.

Thanks Natasha Loder 🐋 for a great conversation on AI’s impact on drug development over the past decade, and why regulators must keep pace to maximize the impact of this new wave of medicines.

Read the full article in The Economist’s Technology Quarterly: economist.com/technology-qua…

Students, there is still a bit of time to sign up for our NeuroAI course! We've got an awesome curriculum for you, courtesy of folks like xaq , Eva Dyer , Nikolaus Kriegeskorte , Surya Ganguli , megan peters 🧠 on bsky & 🐘neuromatch.social , Blake Richards , Grace Lindsay , Guillaume Dumas . Join us!

1/7 So I hear Twitter is now called X.. Guess who's back! 😅 But some things never change... Excited to share our new nature paper reporting a new frequency comb generator - Introducing the FM-OPO microcomb! nature.com/articles/s4158…🌈

Looking forward to giving the Harvard Physics Colloquium this Monday and the MIT Physical Math Seminar this Tuesday, and to catching up with colleagues in Cambridge!

Yes to applications of statistical physics in stats/ML! See eg:

annualreviews.org/doi/abs/10.114…

iopscience.iop.org/article/10.108…

journals.aps.org/prx/abstract/1…

Alexandra Woolgar and I are seeking a postdoc with strong analysis skills! Project applies ML decoding to EEG recordings of autistic non-speakers during audiobook listening, to understand receptive language processing. Based in Cambridge, UK. More details here: tinyurl.com/m9atmjt4

:) discuss!

Absolutely. The fundamentals of these fields will still be fundamental a few decades from now and will need to be learned by those who seek to advance science, technology and society. Curious though, who disagrees and why?