A DbC Inspired Neurosymbolic Layer for Trustworthy Agent Design
This addresses trustworthiness issues in AI agents for developers and users, though it appears incremental as it adapts existing DbC methods to LLMs.
The paper tackles the problem of generative models lacking verifiable guarantees by introducing a contract layer based on Design by Contract principles to mediate LLM calls, resulting in probabilistic remediation to steer generation toward compliance with semantic and type requirements.
Generative models, particularly Large Language Models (LLMs), produce fluent outputs yet lack verifiable guarantees. We adapt Design by Contract (DbC) and type-theoretic principles to introduce a contract layer that mediates every LLM call. Contracts stipulate semantic and type requirements on inputs and outputs, coupled with probabilistic remediation to steer generation toward compliance. The layer exposes the dual view of LLMs as semantic parsers and probabilistic black-box components. Contract satisfaction is probabilistic and semantic validation is operationally defined through programmer-specified conditions on well-typed data structures. More broadly, this work postulates that any two agents satisfying the same contracts are \emph{functionally equivalent} with respect to those contracts.