A Path-Space Formulation of Prediction in World Models: From a Single Action to Prediction, Planning, and Irreversibility
arXiv:2606.28751v1 Announce Type: new Abstract: We propose a path-space formulation of prediction in AI world models. Rather than sequences of one-step conditional distributions, we argue that a world model implicitly defines a probability measure over future trajectories. In the local regime where latent dynamics admit an effective Markovian description, this path measure takes the Onsager-Machlup form. Within this framework, prediction (most probable trajectory), planning (constrained optimization), and uncertainty (fluctuations) emerge as operations on a single action functional. We decompo
This research reflects ongoing efforts within AI to develop more robust and interpretable world models, moving beyond simple sequential predictions.
A path-space formulation of prediction could lead to more efficient and reliable AI agents capable of planning and understanding uncertainty, significantly impacting AI development.
The theoretical framework for how AI world models function could shift from conditional distributions to a more unified action functional, enabling advanced prediction and planning capabilities.
- · AI research institutions
- · Developers of AI agents
- · High-stakes AI applications (e.g., self-driving, robotics)
- · AI models relying solely on one-step predictions
- · Developers using less robust planning heuristics
Improved theoretical understanding of AI world models for prediction and planning.
Acceleration in the development of more capable and autonomous AI agents.
Enhanced trust and deployment of AI in complex, real-world environments due to better predictability and error handling.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG