World Models in Pieces: Structural Certification for General Agents
For AI safety and deployment, this work provides a method to certify the reliability of general agents on specific transitions, enabling trustworthy long-horizon planning.
The paper proves that general agents cannot be universally capable in the big-world regime, rendering standard worst-case analysis uninformative. It introduces structural certification, a transition-local framework that provides entry-wise guarantees on the agent's internal world model, achieving a O(1/n) + O(δ) error bound for deep compositional goals.
In the big-world regime, agents cannot be universally capable and their ability is inevitably specialized across a world model in pieces. Consequently, standard uniform guarantees fail to distinguish between the understanding of critical bottlenecks and irrelevant failures. We first formalize this limitation by proving that general agents are not universal, rendering standard worst-case analysis uninformative. To overcome this, we introduce structural certification, a transition-local framework that maps bounded goal-conditioned performance to entry-wise guarantees on the agent's internal world model. Our main contribution is constructive. We provide algorithms that filter specific transitions using deep compositional goals and prove that a general agent on these goals has a structural world model with a $\mathcal{O}(1/n) + \mathcal{O}(δ)$ error bound. Conversely, this bound is tight in the small-$δ$ regime, whose existence is explicitly guaranteed by our certification. These results enable the certifiable deployment of general agents by localizing the specific transitions where long-horizon planning is reliable.