@mitchellh So far I've found LLM generated release notes are about as good as just posting the raw changelog: the LLMs don't understand the context of a new feature or fixed bug and tend to emphasize everything equally, making it subtly more difficult to parse release notes.
I used to like reading the release notes of projects I don't actively use but planned on some day, since I could quickly scan for that needed feature or bugfix being added, but now they all read the same and blur together in an endless slog of of Bullet point header -> excited description.