I had a very similar mediocre experience using LLMs with Python/Rust/Typescript where I didn’t really see the point. Only when I started using Go for a recent hackathon project did I get the fully hands off experience where I didn’t have to write any code.
I’ve long been on the record saying that type systems are bad and I think this is a big part of what makes Rust/Typescript difficult to track for the LLM.
The problem with Python I think is more that it is too permissive and anything can be anything (and whatever typing we put on top is rather weak and non-standard). I was particularly surprised to notice how much easier it is to read—when working with an LLM I pretty much only read code—Go code than Python code for much of the same reasons.
