Yes in some sense : the ability to have a deep understanding of a language, be it human or computer, is a prerequisite to build and optimize logical blocks like in programming.

The human languages are way more complex than computer languages and they have way more context, requiring to memorize a lot of a text to follow what the subject is.

ChatGPT is really good at discerning what the subject and the intent are while also good a finding related topics.

I would not be surprised if it was revealed that they use similar models to develop, debug or improve their existing code base.

ChatGPT also makes a lot of mistakes which most likely is preventing it to improve without supervision and tweaking from actual developers.
But we are definitely going more and more into what was described as science fiction a few years ago.