The WMF is experimenting with AIs, and that will lead to bad places

[ericbarbour]

Other thread where I noted the "Simple Article Summary" explosion, which only received attention when 404media ran an item about it:

viewtopic.php?p=31317#p31317

It also mentions "Tone Check", which is receiving similar ragey treatment from the Wiki-Inmates.

Is the WMF running OTHER AI projects? Almost certainly. But they are well buried--you can TRY looking thru the subpages of this:
https://research.wikimedia.org/projects.html
It is full of research papers being churned out by "Wikimedia associates" and employees. Good luck parsing the typically impenetrable language of research papers about obscure software subjects. They write like that, partly to to keep the rabble out of their little secret garden of knowledge--exactly the same approach the WP insiders have been taking since its inception, just with greater pretension and slightly more "openness". A word they all love to toss around but never actually respect.

If you know how abusive and crazy WP administrators can often be, this will make you laugh:
https://meta.wikimedia.org/wiki/Researc ... _Attrition
Especially if you've been watching the current attempt to recall Bbb23.

Wanna see some really stupid shit?
https://wikimediafoundation.org/wikipedia-experiments/

Now you know SOME of what happens to the $178 million.

Expect more of this in coming years. Any attempt by the WMF in the past to take "power" away from the lunatic insiders has always been met with incoherent rage, personal attacks on WMF employees, and sockpuppeting and canvassing. Thus does the "wonderful Wikimedia movement" continue its degeneration into a deeply conservative, closed, and paranoid private club. Said it before and saying it again: only a matter of time before WP becomes the "encyclopedia that nobody can edit". Maybe LLM AIs, which were ironically trained on Wikipedia's content, will do a better job of running the damn thing. Maybe not. But you can be certain, the often-insane humans who have been running it will fight like mad dogs to keep their miserable shreds of power.

And the "wiki professionals", generators of impenetrable papers and suckers of cash, will eventually try to push out those often-pathetic amateurs who built the thing, and there will eventually be open warfare.

(Every past attempt to "fork" Wikipedia has failed but there will be more attempts.)

[Strelnikov]

.....And they are going to find out the hard way how badly this version of AI will fail them.

The Phony Comforts of AI Optimism

There Is No AI Revolution

And the "art" and text-barfing is the easy stuff; look at what a battle the self-driving car is!

[journo]

look at what a battle the self-driving car is

Human driven cars are already about 4900% more deadly than, say, for example, rail. They are incredibly, absurdly unsafe by design, AI can't help something so fundamentally flawed in the next few hundred years. If the speed limit is reduced to something sensible, it could be automated somewhat safely, but it would seem better to drop our car obsession.

Wikipedia would also certainly need to look at its structural problems or whether it is a net societal benefit before automating itself. Has society improved since they became the center for Anglo-Saxon knowledge 15+ years ago?