I wonder about using "failscene" to describe the currently-popular slate of AI tools and demos. In contrast with the demoscene, which is about getting very low powered computers to do cool things you wouldn't 11 minutes of HD video in 64 kilobytes expect them to be able to do, the failscene is about getting very high powered computers to fail at doing uninspiring things we already know how to do without them. You can even stylize it fAIlscene if you're so inclined.
Some folks in the field say that practically speaking you can't get anywhere with your research without a killer demo. What they seem to miss is what happens when your field becomes all demos all the way down.
