[ExI] Arxiv and AI slop

[Ben Zaiboc]

On 11/04/2026 16:23, John K Clark wrote:
>
> On Fri, Apr 10, 2026 at 5:25 AM BillK via extropy-chat <extropy-chat at lists.extropy.org> wrote:
>
>     > One response is to fight fire with fire by using AI in peer review or to weed out fake papers.
>
>
> I'm not sure I understand how that could work. It seems to me the important thing is not in determining if a paper was written by an AI or a human, it's in determining if the paper is any good; or when it comes to stuff that has not yet been peer reviewed, at least a 10% chance that the paper will turn out to be pretty good. I don't suppose it would take a very advanced AI to weed out papers that contained obvious logical blunders, or those that go on and on about the properties of magical crystals from Atlantis and other such nonsense. 


I'm not sure how /that/ could work.

As far as I'm aware, we still haven't solved the problem of AI systems just making stuff up, so how could we be sure that the 'weeding out' process actually did weed out logical blunders and various nonsense?

-- 
Ben