"Achieving that goal necessitates a production system supporting it" is very different from "If the control system is a full team in a remote location, this vehicle is not autonomous at all" which was what GP said.
LLMs are nothing like compilers. This sort of analogy based verbal reasoning is flimsy, and I understand why it correlates with projecting intelligence onto LLM output.
Oh a working language learning app? Like one of the hundreds that have been shown on HN in the past 3 years? But only demonstrated to be some generic single word translation game?