NEW PAPER 🧵: Deep neural networks are complicated, but looking inside them ought to be simple. In this paper we introduce TorchLens, a package for extracting all activations and metadata from any #PyTorch model, and visualizing its structure, in as little as one line of code. 1/

Thrilled to share that I'll be joining Harvard Medical School’s Dept. of Neurobiology & the Kempner Institute at Harvard University as a founding faculty member in September 2023: harvard.edu/kempner-instit…

Absolutely thrilled to share my postdoc work in the Axel lab. We found odor-evoked representations of the intrinsic value of information in mouse orbitofrontal cortex and showed that mice desire knowledge as its own reward. Now on bioRxiv! a #tweeprint 🧵: biorxiv.org/content/10.110…

Next we created a nonlinear latent variable model of OFC activity in our task using CEBRA, the awesome new method from Mackenzie Mathis, PhD' lab, to understand how the task is encoded in the neural circuit at the level of aggregate neural dynamics

The separate water and information value representations were also evident in low-dimensional embedded dynamics in OFC, suggesting they are an important decodable aspect of OFC activity during information seeking.

What is scientist Jennifer J Bussell @jenbussell.bsky.social curious about? Curiosity itself! This postdoc’s research in the Axel lab explores the origins of the brain’s hunger for information. #brain #learning #curiosity youtu.be/_3KI3y90lcU

Great video on our ongoing collab with Jennifer J Bussell @jenbussell.bsky.social Columbia University's Zuckerman Institute looking at the brain's value of info as its own reward & the underlying neural mechanisms. Congrats Ryan P. Badman! First paper from Rajan lab's new home Harvard Medical School Kempner Institute at Harvard University. Read: biorxiv.org/content/10.110…

Boston autumn pictures from the walk to work.

Our paper explores the risks of using LLMs for strategic military decision-making. With OpenAI quietly updating their ToS to no longer prohibit military and warfare use cases, this research is more crucial than ever. These models aren’t safe to use in high-stakes situations yet.

A major challenge for computational social neuroscience is to go to more naturalistic small groups, while still doing controlled, goal-oriented experiment & analysis. We present a strong effort in that direction (from RIKEN). Fig format issues fixed, v2: biorxiv.org/content/10.110…

New survey on predictive representations in reinforcement learning, covering AI, cognitive, and neuroscience perspectives: arxiv.org/abs/2402.06590 Joint work with Wilka Carvalho, Momchil Tomov, Will de Cothi, Caswell Barry

Now out in @nature: Artificial Intelligence and Illusions of Understanding in Scientific Research, coauthored with the brilliant @lisamesseri and edited by the supremely talented Mary Elizabeth rdcu.be/dAv39🧵

New paper published by @ICMLconf #icml2024 Riley Simmons-Edler, Ryan P. Badman, Shayne Longpre & I argue that global embrace of military #AI systems with less human oversight makes conflict more likely, jeopardizing global stability & AI research norms: arxiv.org/abs/2405.01859

We show how committed global militaries have become to AI weapon development & deployment, and: 1) Offer potential responses for academic institutions, the AI research community & governing bodies 2) Highlight the importance of caution & transparency going forward

🚨 New #ICML2024 position piece. The most overlooked risks of AI stem from autonomous weaponry For 4 reasons: 1⃣ Arms race w/ ⬇️ human oversight 2⃣ Reduces cost of starting conflicts 3⃣ Evades accountability 4⃣ Battlefield errors aren’t considered costly See our work led by

🔭 New perspective piece at #ICML2024 & Massachusetts Institute of Technology (MIT) AI Impact Award winner🎉 🌟Data Authenticity, Consent, and Provenance for AI Are All Broken: What Will It Take to Fix Them?🌟 w/ Robert Mahari Naana Obeng-Marnu Will Brannon Tobin South katy ilonka gero Alex Pentland Jad Kabbara 🔗:

Thank you Catherine Caruso for this thoughtful Q&A on our recent paper by Ryan P. Badman Riley Simmons-Edler & Shayne Longpre about using AI in military tech. It’s been amazing to see our community's reception to work beyond our usual #neurotheory papers. More: hms.harvard.edu/news/risks-art…

Happy to deliver a keynote address on military attitudes of trust in AI at this conference in Montreal, Canada from 02-03 December. Sign-up to share your own research! Cornell Brooks Tech Policy Institute (BTPI) Mila - Institut québécois d'IA Ryan P. Badman Harvard Medical School U.S. Army War College Army Strategist harms-risks-ai-military.org/speakers

Grateful 🌟Consent in Crisis: The Rapid Decline of the AI Data Commons🌟 was accepted to NeurIPS Conference 2024! This was a massive undertaking to audit 14k+ websites underneath popular pretraining corpora. See you in Vancouver!