General agents need world models

Jonathan Richens, Tom Everitt, David Abel
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:51659-51687, 2025.

Abstract

Are world models a necessary ingredient for flexible, goal-directed behaviour, or is model-free learning sufficient? We provide a formal answer to this question, showing that any agent capable of generalizing to multi-step goal-directed tasks must have learned a predictive model of its environment. We show that this model can be extracted from the agent’s policy, and that increasing the agents performance or the complexity of the goals it can achieve requires learning increasingly accurate world models. This has a number of consequences: from developing safe and general agents, to bounding agent capabilities in complex environments, and providing new algorithms for eliciting world models from agents.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-richens25a, title = {General agents need world models}, author = {Richens, Jonathan and Everitt, Tom and Abel, David}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {51659--51687}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/richens25a/richens25a.pdf}, url = {https://proceedings.mlr.press/v267/richens25a.html}, abstract = {Are world models a necessary ingredient for flexible, goal-directed behaviour, or is model-free learning sufficient? We provide a formal answer to this question, showing that any agent capable of generalizing to multi-step goal-directed tasks must have learned a predictive model of its environment. We show that this model can be extracted from the agent’s policy, and that increasing the agents performance or the complexity of the goals it can achieve requires learning increasingly accurate world models. This has a number of consequences: from developing safe and general agents, to bounding agent capabilities in complex environments, and providing new algorithms for eliciting world models from agents.} }
Endnote
%0 Conference Paper %T General agents need world models %A Jonathan Richens %A Tom Everitt %A David Abel %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-richens25a %I PMLR %P 51659--51687 %U https://proceedings.mlr.press/v267/richens25a.html %V 267 %X Are world models a necessary ingredient for flexible, goal-directed behaviour, or is model-free learning sufficient? We provide a formal answer to this question, showing that any agent capable of generalizing to multi-step goal-directed tasks must have learned a predictive model of its environment. We show that this model can be extracted from the agent’s policy, and that increasing the agents performance or the complexity of the goals it can achieve requires learning increasingly accurate world models. This has a number of consequences: from developing safe and general agents, to bounding agent capabilities in complex environments, and providing new algorithms for eliciting world models from agents.
APA
Richens, J., Everitt, T. & Abel, D.. (2025). General agents need world models. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:51659-51687 Available from https://proceedings.mlr.press/v267/richens25a.html.

Related Material