maaaaybe these unnamed instances are using this to root out reactionary content? (still silly, AI gives you yes or no answers basically on an overcomplicated coinflip and spits out text to justify that initial answer) But c'mon. Freaking out about "people we strongly disagree with" using a sketchy moderation tool on "their instances"?

A specter is haunting the fediverse, and that spectre is hexbear-specter . Well really its mostly lemmy.ml but sometimes they remember we exist too.

It sounds like they're using this to find "tankie" users and they're scared that the "tankie" instances will use this (stupid, counterproductive) moderation style on them (lol). Admittedly conjecture on my part.

screenshot from piefed.social: AI-assisted moderation in the fediverse is happening. Now what? by Rimu. I recently discovered that some popular federated instances have been using LLM-assisted moderation tooling that evaluates whether someone has said something bannable. They do this by running a script/app that sends the user’s comment history to OpenAI with the question “analyze this content for evidence of specific political ideology sentiment. Also identify any related political ideology tropes“. (The italic bits are where I’ve redacted the ideology they’re seeking).

you are viewing a single comment's thread
view the rest of the comments

[–] ZWQbpkzl@hexbear.net 12 points 4 months ago

When you actually dig into the comments, its not whole instances, its just one mod, actually its the other db0 admin, Unruffled, who recently banned the OP, whose also the piefed author, Rimu.

catgirl-flop

Imagine getting this far up your own ass about the fediverse.

Fwiw, the db0 admins seem to know enough about AI that they could be doing this all locally with a 4B model. Forwarding all that content to a data center model would be just disgraceful.

permalink

fedilink

source