Scholar Conversation · Artificial Intelligence
“Nobody needs to know what the system usually does”
Dr. Aarav Mehta on formal verification, the value of a negative result, and what an industrial partner taught him about his own assumptions.
Dr. Aarav Mehta completed his doctorate at Oxford in 2026 on the formal verification of learned controllers, working in collaboration with an industrial partner operating process-control systems. He spoke to us shortly after his profile was published.
Corrections? Contact the desk.
Continue reading
All writing
More on artificial intelligence
Research Summary
What interpretability actually requires
A decade of explanation methods has produced impressive demonstrations and remarkably few guarantees. A reading of the formal literature, and of what…
Artificial Intelligence Sep 3, 2026 9 min read