Maybe society has a bias and the source data fed into LLMs simply reflects that. Or maybe, if you're far enough to the right, nearly everything else *looks like* left-wing bias even though much of it is the mainstream view. - ThreadSky

About ThreadSky

ndixon.bsky.social • 459 days ago

Maybe society has a bias and the source data fed into LLMs simply reflects that.

Or maybe, if you're far enough to the right, nearly everything else *looks like* left-wing bias even though much of it is the mainstream view.

Comments

matrix07012.bsky.social•459 days ago

Yes there's a bias from the training data, that will always be there, but ChatGPT has an additional model stopping it from generating certain things and has prebaked responses on some topics.

ndixon.bsky.social•458 days ago

Yeah but the only people who want to see those blocked responses are cunts.

matrix07012.bsky.social•457 days ago

matrix07012.bsky.social•440 days ago

Here is what Microsoft uses and it considers statement of fact toxic and has a straight up anti white bias.
https://arxiv.org/abs/2203.09509

ndixon.bsky.social•438 days ago

Did you read and understand all of that study, or just swallow some hot take on Rumble?

It shows current automated tools get it wrong, but accuracy can improve by tweaking the parameters

So some subtle racism that went under the radar can now be flagged, which may make some of you uncomfortable

Posting Rules

Be respectful to others
No spam or self-promotion
Stay on topic
Follow Bluesky's terms of service

Comments

Posting Rules

Reply