How do we infer the goals & plans of others from both their actions & words? In this paper with @Lance_Ying42, we infer a team's goal via inverse planning (aka "inverse RL"), using LMs* as likelihood functions over utterances! (*GPT-3 Curie 6.7B, but smaller LMs may also work!)

I wrote a thought piece showing RLHF = variational inference on Bayesian cognitive model (generalized RSA). I hope that realizing this connection can help better understand recent developments on LLMs and inspire future research. arxiv.org/abs/2305.17760

👋Are you excited to meet other researchers working on theory-of-mind? We’re organizing a 𝒎𝒆𝒆𝒕-𝒂𝒏𝒅-𝒈𝒓𝒆𝒆𝒕 session🤩📢 to match people with shared interests. Open to both in-person and virtual!! 🎯Sign up at forms.gle/xbx6wMASgYdXED… See y'all at #ICML2023 on 7/28🌅

Bayesian theory of mind (BTOM) has mostly been used to model how humans understand each other's mental states and biases (e.g. inaccurate world model). Interestingly, we found that BTOM tends to make pessimistic inference of other agents' beliefs. (1/n)

How does deception, skepticism, and paranoia emerge from Theory of Mind? Excited to present TWO orals on this tomorrow at the ToM workshop at #ICML2023 ToM Workshop, both with the amazing nitay alon. Brief thread below 👇

New work from us - reviewed & accepted to ToM Workshop Our new theoretical work outlines when deep mentalisation is useful, what happens when it's used outside of appropriate contexts, & how it might give rise to psychopathology thread from talented colead author nitay alon👇

Are you at ICML? Come hear about SymbolicToM at ToM Workshop 🧠: oral presentation 14:25 HST, poster session 12:30 HST! Thrilled to chat with everyone about all things theory of mind, DM me to chat 1:1!

What does it mean for AI to be socially competent? 🧐 In our recent work (with juscelina portela, Peter Ford Dominey, and Pierre-Yves Oudeyer ) we explore those topics (to be presented at ToM Workshop today). arxiv.org/abs/2307.07871

In ICML? --> Check out the Theory of Mind workshop! ToM Workshop In the Theory of Mind workshop? --> Check out Sonia Murthy's Sonia Murthy @ COLM work! "Comparing the Evaluation and Production of Loophole Behavior in Children and Large Language Models" 🤖🧒➰

Looking forward to be part of the panel discussion ToM Workshop sharing thoughts and learning about interactive agents and environments involving social utilities and strategic communication. HST 4-5 PM! #ICML2023

Presenting two papers at the ToM Workshop poster session from 12.15-13.30 today! Come by if you're interested in how to use LLMs as sub-components of larger Bayesian models. 1. Inferring the Goals of Communicating Agents from Actions and Instructions arxiv.org/abs/2306.16207

Starting now! ToM Workshop

📢 ToM workshop @ ICML is starting now with Maarten Sap (he/him)'s talk!! 😍 Come and join us at 317A Kaho'olawe Room 🏝️😎

Excited to present our ongoing work investigating loophole behavior in LLMs today at ToM Workshop! #ICML2023 #NLProc

Poster Session at ToM 2023 is happening! Room 317A check out if you are at ICML

Come to Room 317A to check out our awesome list of posters!

We are honored to have Jesse Thomason Bodhisattwa Majumder Kevin McKee joining our discussion panel from 4-5pm HST. Please submit questions you want them to discuss through app.sli.do/event/mZXwiZsk…

Excited to be presenting a poster comparing GPT-3 and humans on a set of 6 Theory of Mind Tasks that we call 'EPITOME'. GPT-3 matched humans on 3 tasks, but struggled with those that require pragmatic reasoning. ICML Conference ToM Workshop openreview.net/forum?id=e5Yky… w/ Sean Trott

Everything I learned at the first Workshop on Theory of Mind in Communicating Agents ToM Workshop ICML Conference: Here I’ll cover some of the main takeaways from the keynotes, oral presentations, and posters from the workshop (not exhaustive) 🧵 #ICML2023