LLM-generated Rust similarities

In our team we’ve recently noticed that, whenever any agent generates Rust, it always tends to be in the same specific style. It’s very evident, like non-idiomatic, very much imperative style, especially noticeable after coming from Scala into Rust (it even generates Scala in the same style! A lot of vars, while loops etc.). It’s as if it always tries to follow a very impure way of programming, so to speak. Either in existing code base or in a new one, always the similar result.

Has anyone also noticed this? It’s as if a language model was allergic to a more declarative programming style.