From monoliths to modules: Decomposing transducers for efficient world modelling

TL;DR

This paper introduces a framework for decomposing complex transducer-based world models into sub-transducers, enabling more efficient, interpretable, and distributed inference for AI safety and real-world applications.

Contribution

It develops a novel method for inverting transducer composition to derive sub-transducers, facilitating modular, parallelizable, and transparent world modeling.

Findings

Derived sub-transducers operate on distinct input-output subspaces.

Enables parallelizable and interpretable world models.

Supports distributed inference for efficient real-world AI applications.

Abstract

World models have been recently proposed as sandbox environments in which AI agents can be trained and evaluated before deployment. While realistic world models often have high computational demands, this can often be alleviated by exploiting the fact that real-world scenarios tend to involve subcomponents that interact in a modular manner. In this paper, we explore this idea by developing a framework for decomposing complex world models represented by transducers, a class of models generalising POMDPs. Whereas the composition of transducers is well understood, our results clarify how to invert this process by deriving sub-transducers operating on distinct input-output subspaces, enabling parallelizable and interpretable alternatives to monolithic world modelling that can support distributed inference. Overall, these results lay groundwork for bridging the computational efficiency required for real-world inference and the structural transparency demanded by AI safety.