News for hackers. Stuff that matters.

A comment on The note Reddit sent to moderators threatening them if they don’t reopen. Parent comment

Re:The note Reddit sent to moderators threatening them if th... by ribosometronome on Friday June 16, 2023 @01:05PM (#36360081)
I wonder what dang thinks about that for HN?

I'm not sure how you would properly prompted LLM change moderating standards based on community feedback? Sticky threads for popular topics? Have the understanding of things like local politics or a newly released game in 2023 to be able to separate out bad faith discussion from real? Reddit isn't twitter where the main thing the mods do is remove abusive replies. They're community crafters and maintainers.

[ Parent | on HN ]
Re:The note Reddit sent to moderators threatening them if th... by SanderNL on Friday June 16, 2023 @01:36PM (#36360642)
Alright, I see. Reddit is not my thing, but I can see how this gets complex (and costly) fast.
[ Parent | on HN ]
Re:The note Reddit sent to moderators threatening them if th... by seanhunter on Friday June 16, 2023 @02:28PM (#36361453)
My understanding is Reddit the company actually already uses openai's moderation api[1]. Not sure they expose that to anyone (even mods) outside the company. It's not an LLM I don't think though, it's just a set of classifiers to give a probability that something violates various categories of their TOS.

[1] https://platform.openai.com/docs/guides/moderation

[ Parent | on HN ]
Re:The note Reddit sent to moderators threatening them if th... by 1024core on Friday June 16, 2023 @02:12PM (#36361157)
A few-shot prompt should be able to do it. Now, I will admit I have not tried; but you could have the prompt be some posts that are considered "on topic" (of the sub) and others that are "off topic", and let the LLM rate these posts. Limit the number of new posts to a small number to prevent probing and spamming.
[ Parent | on HN ]
Re:The note Reddit sent to moderators threatening them if th... by ribosometronome on Friday June 16, 2023 @02:56PM (#36361941)
And handling the 11th unique story on a single topic being posted and voted up by brigades that week? Considering people's user history when making moderation judgments? Having the context to understand new dog whistles? All things that are technically possible, I guess, but it requires a lot of complex prompting, long context windows, and effort placed into moderation tools that Reddit doesn't seem to have put significant effort into before.
[ Parent | on HN ]