Thanks for the feedback! And indeed, I do agree!
Practically for this jam I ran out of time, my original vision was to not use gen AI for dialogue or story content, but when I was looking down at the barrel at 3h left in submission and most still being unwritten, I just put it through gen AI machine to fill my fluent keys. It was a choice between not publishing, publishing something much less or using non-curated AI to write me stuff (I had written a story bible so it helped a bit). I chose the latter, but there were no good choices. I overscoped and failed in that :D
Actually, I managed to make the critical bug fixes a few minutes past the jam game, and I continued the project immediately on here:
https://lazyanttu.itch.io/the-unstable-key
That now has about 60-70% of the content written by me (I have replaced the majority of the AI-written stuff), with lots of bug fixes, and is much better. Of course, the rating is based on what I submitted on the jam, but I wanted to share some post-jam progress! :)
That was helpful to confirm my suspicion that it was a good idea to continue replacing the prose. I also think the issue is that there is so much variance in quality; I wrote some segments of the game carefully, and then there is a lot of slop. So those kinds of conflict with each other.
So the plan to use gen AI to write was not plan A; it was plan B once I ran out of time.