Live data from Hacker News

AI Slop Is Flooding Medium

wired.com

1–10 of 76 posts

Re: AI Slop Is Flooding Medium

#5
Previously, spam has been reported and filtered via "flag" and "report" buttons.

Why is that not sufficient for site operators to separate "AI Slop" from other articles?

If existing sites choose to tolerate "AI Slop" (why doesn't HN have this problem?), does that create a market opportunity for competitors with better filters?

Re: AI Slop Is Flooding Medium

#6
YouTube allows one to report a video for 11 different reasons. A list below. Unfortunately, AI Slop is not one of them.

Sexual content

Violent or repulsive content

Hateful or abusive content

Harassment or bullying

Harmful or dangerous acts

Misinformation

Child abuse

Promotes terrorism

Spam or misleading

Legal issue

Captions issue

Re: AI Slop Is Flooding Medium

#7

Previously, spam has been reported and filtered via "flag" and "report" buttons. Why is that not sufficient for site operators to separate "AI Slop" from other articles? If existing sites choose to tolerate "AI Slop" (why doesn't HN have this problem?), does that create a market opportunity for competitors with better filters?

Before LLMs, the cost to generate spam content was non-zero. Now, it's effectively zero. These infinite supplies of AI Slop should be handled differently and more aggressively that legacy spam.

Re: AI Slop Is Flooding Medium

#8

Previously, spam has been reported and filtered via "flag" and "report" buttons. Why is that not sufficient for site operators to separate "AI Slop" from other articles? If existing sites choose to tolerate "AI Slop" (why doesn't HN have this problem?), does that create a market opportunity for competitors with better filters?

Mostly because most spam isn't going to be reported. Traditional anti-spam works because once you have a handful of reports, you can probably generalize it to an entire class of spammy pages generated using a particular technique, and get rid of everything.

LLM-generated content is not distinguishable from human content by any simple rule, so you lose the ability to police the platform that way.

Re: AI Slop Is Flooding Medium

#10
post #6

YouTube allows one to report a video for 11 different reasons. A list below. Unfortunately, AI Slop is not one of them. Sexual content Violent or repulsive content Hateful or abusive content Harassment or bullying Harmful or dangerous acts Misinformation Child abuse Promotes terrorism Spam or misleading Legal issue Captions issue

Why should it be one of them? It's either spam or not a problem beyond “I find it distasteful.”
Post reply on HN