English Wikipedia, mainspace.
1,000People are also trading
My experience with AI topics is that "techbros" really overrate the competence of AI consistently, especially when the problem statement has nothing to do with AI.
In this case, this is very much a Wikipedia question. So even if o4 and o5/o6 make massive enough leaps that they mimic a Wikipedia article well enough, I consider an approximately <0.1% chance it will be accepted by the English Wikipedia community at large
@xxx I will add that, in the spirit of the question, if it evades reversion merely because it just so happens that nobody sees it, because it's obscure, or something like that, rather than because of the quality of the writing, I would not consider that to be a success.
@BenjaminIkuta what if just barely happens on a technicality, like grammar correction? Because Wikipedia has a lot of mundane tasks where setting up the ai is more work than doing it.
@BenjaminIkuta Wikipedia editing is a polticial process dominated by nerd fascism. Not even humans can edit Wikipedia without heavy review by moderators so unless Wikipedia symbolically allows an ai to make a token edit it will never be allowed.
@EduardoFilippi Tables of sports scores aren't exactly "substantive original cited prose."
When you say 'unassisted', is it OK for a human to prompt the AI by asking/triggering a command to make it edit a particular page, or does the AI have to be fully autonomous? In the latter case, regardless of capabilities, wikipedia might well ban such activity (or in the former, but it's somewhat less likely).
@AngolaMaldives If the human can review the edit and choose to stop it before it's published, then it doesn't count.