Learning-Infused Formal Reasoning: From Contract Synthesis to Artifact Reuse and Formal Semantics

The paper presents a vision for integrating formal methods with artificial intelligence, focusing on automated contract synthesis, semantic artifact reuse, and refinement-based theory. It proposes a hybrid framework that combines large language models with graph-based representations to enhance semantic matching and reuse of verification artifacts, ensuring formal soundness through symbolic matching. This approach aims to create verification ecosystems that leverage cumulative knowledge for more efficient and scalable correctness proofs, which is crucial for practitioners working on formal verification and AI integration.