Formal verification is a technique that uses mathematical methods to prove code correctness. Despite its long history, it has been confined to research fields because writing proofs is extremely difficult and time-consuming. Author Martin Kleppmann predicts that AI assistants based on large language models (LLMs) will completely transform this situation. AI can now help automate the writing of proof scripts, making formal verification affordable and efficient. This will make verification more feasible, while AI-generated code will also require formal verification to ensure correctness, replacing manual review. In the future, developers will only need to specify desired properties, and AI will generate both code and proofs, similar to how a compiler works. This shift will make formal verification a mainstream practice in software development, although challenges will shift to correctly defining specifications. AI’s precision can also compensate for the uncertainties of large language models, improving overall reliability. This transformation heralds a new era in software engineering, where cultural adaptation will be key.
