Runtime Verification is hosting a live session on formal verification for AI-generated code and safer AI agents.
Speakers:
• Grigore Rosu (Founder, Runtime Verification) - 9:05 AM PT
• Everett Hildenbrandt (CEO, Runtime Verification) - 9:30 AM PT
The discussion centers on applying formal specs to AI-written code. As LLMs generate more production code, runtime verification becomes critical for catching edge cases and ensuring correctness that unit tests miss. Rosu pioneered the K Framework for executable formal semantics, which has been used to verify smart contracts and VM implementations.
Key angle: AI agents need provable guarantees, not just probabilistic outputs. Formal methods can bridge the gap between LLM flexibility and system reliability requirements. This matters for autonomous agents handling high-stakes operations where "mostly correct" isn't acceptable.
Speakers:
• Grigore Rosu (Founder, Runtime Verification) - 9:05 AM PT
• Everett Hildenbrandt (CEO, Runtime Verification) - 9:30 AM PT
The discussion centers on applying formal specs to AI-written code. As LLMs generate more production code, runtime verification becomes critical for catching edge cases and ensuring correctness that unit tests miss. Rosu pioneered the K Framework for executable formal semantics, which has been used to verify smart contracts and VM implementations.
Key angle: AI agents need provable guarantees, not just probabilistic outputs. Formal methods can bridge the gap between LLM flexibility and system reliability requirements. This matters for autonomous agents handling high-stakes operations where "mostly correct" isn't acceptable.