News for hackers. Stuff that matters.

neuropsha

Re:Clarkesworld AI Submissions Update on Thursday March 23, 2023 @01:16PM (#35277995)
I don’t know why you’ve been downvoted. You are, sadly, right. Being a reader for a publication or literary agent was already a gruelling job before AI made the slush problem a hundred times worse. They don’t go into the role wanting to fire off form rejections, but there is too much hopeless perma-slush for them to do much else.

We’re going to end up with AI autograders (for quality as much as provenance) as a first pass before a human read. On the other hand, if the AIs are selective enough, they can guarantee that each passing submission can be afforded a serious, deep read… which isn’t the case right now unfortunately.

Re:Clarkesworld AI Submissions Update on Thursday March 23, 2023 @12:49PM (#35277554)
And because, inevitably, there is someone in the crowd who is blinded by the tech and screaming “judge it for the quality” … No, that would be a waste of time.

This essay, which is the most in-depth analysis of this problem I’ve found so far, suggests that we’ll have to find an automated way to filter for quality, because AI-generated content is only going to get more sophisticated: https://antipodes.substack.com/p/ai-will-kill-literature-and...

The endgame seems to be the widespread deployment of AI autograders that, while they won’t be able to judge literary merit, will at least be as good as the current, mostly social, signals used today.