Comment by shchegrikovich

I agree; we can do much more with LLMs if we focus on declarative-style languages.