I get what you mean but I think if anything AI pairs extremely well with strongly typed languages that are at times cumbersome for humans, but decrease the latency at which AI can get feedback on its code. In my (very) limited experience Rust is an excellent target for AI codegen.

Clojure is a strongly typed language. A Clojure REPL capturing immutable, inspectable state is a philosophically richer feedback substrate than it gets credit for. Spec can express constraints that static types cannot - things that would require dependent or refinement types in a static system (and the enormous complexity that comes with them) you can just write as a predicate. The tradeoff is that specs are only checked when you actually exercise the code path, whereas a type error is total and upfront. But that's exactly the point - an agent working surgically on a specific path is exercising it, so the totality of static checking matters less. If we're not vibe-coding, a dynamic, strongly-typed, immutability-friendly language like Clojure could be both token-efficient and capable of richer reasoning than a static type system allows.