News for hackers. Stuff that matters.
Re:Why Google's 'woke' AI problem won't be an easy fix by chasd00 on Wednesday February 28, 2024 @12:19PM (#39540948)
I wonder how deep in the model the bias is. If it's not a matter of adjusting a system prompt or minor fine tuning it could be a while before there's another release. I know one technique of de-censoring a model is basically fine tuning out the censorship so maybe that's what they're doing to Gemini.

There are no comments on this comment yet.