Wolfgang Lehrach | Code World Models for General Game Playing | Stanford HAI
Stanford
University
  • Stanford Home
  • Maps & Directions
  • Search Stanford
  • Emergency Info
  • Terms of Use
  • Privacy
  • Copyright
  • Trademarks
  • Non-Discrimination
  • Accessibility
© Stanford University.  Stanford, California 94305.
Skip to content
  • About

    • About
    • People
    • Get Involved with HAI
    • Support HAI
    • Subscribe to Email
  • Research

    • Research
    • Fellowship Programs
    • Grants
    • Student Affinity Groups
    • Centers & Labs
    • Research Publications
    • Research Partners
  • Education

    • Education
    • Executive and Professional Education
    • Government and Policymakers
    • K-12
    • Stanford Students
  • Policy

    • Policy
    • Policy Publications
    • Policymaker Education
    • Student Opportunities
  • AI Index

    • AI Index
    • AI Index Report
    • Global Vibrancy Tool
    • People
  • News
  • Events
  • Industry
  • Centers & Labs
Navigate
  • About
  • Events
  • AI Glossary
  • Careers
  • Search
Participate
  • Get Involved
  • Support HAI
  • Contact Us

Stay Up To Date

Get the latest news, advances in research, policy work, and education program updates from HAI in your inbox weekly.

Sign Up For Latest News

Your browser does not support the video tag.
eventSeminar

Wolfgang Lehrach | Code World Models for General Game Playing

Status
Upcoming
Date
Wednesday, May 13, 2026 12:00 PM - 1:15 PM PST/PDT
Location
353 Jane Stanford Way, Stanford, CA, 94305 | Room 119
Topics
Machine Learning
Natural Language Processing
Attend Virtually

While Large Language Models (LLMs) show promise in many domains, relying on them for direct policy generation in games often results in illegal moves and poor strategic play.

Share
Link copied to clipboard!
Event Contact
Stanford HAI
stanford-hai@stanford.edu

Related Events

Ashesh Rambachan | From Next-Token Prediction to Automatic Induction of Automata
Apr 13, 202612:00 PM - 1:00 PM
April
13
2026

Sequence data is ubiquitous in economics — job histories in labor economics, diagnosis and treatment sequences in health economics, strategic interactions in game theory. Generative sequence models can learn to predict these sequences well, but their complexity makes it hard to extract interpretable economic insights from their predictions.

Event

Ashesh Rambachan | From Next-Token Prediction to Automatic Induction of Automata

Apr 13, 202612:00 PM - 1:00 PM

Sequence data is ubiquitous in economics — job histories in labor economics, diagnosis and treatment sequences in health economics, strategic interactions in game theory. Generative sequence models can learn to predict these sequences well, but their complexity makes it hard to extract interpretable economic insights from their predictions.

Caroline Meinhardt, Thomas Mullaney, Juan N. Pava, and Diyi Yang | How Can AI Support Language Digitization and Digital Inclusion?
SeminarApr 15, 202612:00 PM - 1:15 PM
April
15
2026

What does digital inclusion look like in the age of AI? Over 6,000 of the world’s 7,000-plus living languages remain digitally disadvantaged.

Seminar

Caroline Meinhardt, Thomas Mullaney, Juan N. Pava, and Diyi Yang | How Can AI Support Language Digitization and Digital Inclusion?

Apr 15, 202612:00 PM - 1:15 PM

What does digital inclusion look like in the age of AI? Over 6,000 of the world’s 7,000-plus living languages remain digitally disadvantaged.

Matt Beane | Precision Proactivity: Measuring Cognitive Load in Real-World AI-Assisted Work
Apr 20, 202612:00 PM - 1:00 PM
April
20
2026

Systems like ChatGPT and Claude assist billions through proactive dialogue—offering unsolicited, task-relevant information. Drawing on Cognitive Load Theory, we study how cognitive load shapes performance in AI assisted knowledge work.

Event

Matt Beane | Precision Proactivity: Measuring Cognitive Load in Real-World AI-Assisted Work

Apr 20, 202612:00 PM - 1:00 PM

Systems like ChatGPT and Claude assist billions through proactive dialogue—offering unsolicited, task-relevant information. Drawing on Cognitive Load Theory, we study how cognitive load shapes performance in AI assisted knowledge work.

In this talk, I present an approach that moves away from direct prompting, instead using LLMs as program synthesizers to bridge the gap between natural language rules and symbolic world models.  The LLM receives a game description and example trajectories, and outputs an executable, symbolic world model (CWM) represented in Python. The trajectories also ensure the rules are correctly captured and aid in refining the CWM if they are not.  Note that even trajectories containing only a single player's observations and actions can be used to help validate and refine CWMs.  Furthermore, partially observed trajectories also allow comparisons between CWMs via a bound on the likelihood.  


Given a CWM, Monte Carlo Tree Search (MCTS) or Reinforcement Learning (RL) methods can play the game, and gameplay can be further enhanced by adding in LLM-derived synthesized value functions. Imperfect information games are handled by having the LLM synthesize inference functions to impute information sets, or by directly training reinforcement learning policies on top of the CWM.

Speaker
Wolfgang Lehrach
Staff Research Scientist, DeepMind