But you did not publish them on Flathub either, right? The assumption is, that published tools have reached a certain maturity that is meant to be used by public. So if people don't use Ai, then they don't even publish unfinished projects they won't work on. I also don't need to use Ai to make mistakes. That does not invalidate the data and arguments made in this research results.

Yeah I actually liken it to that too. There have always been lots of one-off throwaway projects out there and that's fine. These little one-off slop projects are fine too, as long as everyone understands what they are and they're not pushed in anyone's face or polluting any digital shelves where people are looking for real products. It's no different than automating something with a shitty little forgettable bash script. It's fine for personal use, you can of course freely share it with someone who's clearly struggling with the same issue themselves, but it doesn't really merit advertising it as an awesome new tool for everyone.

One of the (many) big problems with AI is how convincing it can make things appear. A shitty little throwaway bash script is obviously a throwaway bash script and nobody's going to mistake it for quality software, maybe it does the job fine in a very narrow window of circumstances but it's not safe to roll out widely. A throwaway AI project on the other hand can appear like it's a fucking well-documented production-quality scalable enterprise solution in a box. It's NOT. But the fact that it convinces people it is (both the author and the users) is the biggest problem with this. It's all fake. It's full of bugs. The documentation is nonsense. Half the features don't even work, or are literally just fake UI boilerplate that looks like it works until you realize it was never really doing anything at all just failing behind the scenes and possibly deleting or corrupting your entire production database while cheerfully creating the illusion of success.

AI is based on an interesting technology but at least the way it's being deployed today it was not designed to be correct, even if sometimes it is, it was designed to be manipulative, sycophantic and convincing and that's why it's leading people into AI psychosis and creating all these slop-peddlers who legitimately think they are changing the world with their sloppy little throwaway projects. AI is being used and developed as a hack against the human brain and human society. We are fallible, gullible creatures, and they are trying to use it to scour our entire economy for money by conning us into believing it's more than it is. It won't work, but its convinced them it will, and it's convincing us it will, and it was even sort of working for awhile. It won't work anymore once the bubble pops, but we're apparently not quite there yet.