Re:Why Google's 'woke' AI problem won't be an easy fix
by chasd00 on Wednesday February 28, 2024 @01:19PM
(#39541750)
The should just add a DEI slider with values from 0 to 100 and initialized to 50 and adjust the system prompt based on what the user selects. If someone screams to high heaven that gemini is too woke then reply with their DEI slider setting and say "well you had it turned up to 100". If someone screams Gemini is biased and unfair then the reply would be "you turned the DEI slider to 0".
Make the user be the one responsible for the behavior.
They were trying to combat bias from training that has hurt them in the past. The infamous, "all doctors are evidently white american men." To do that, they tried messing with the prompts that users gave, and this led to the craziness they are currently facing.
They would like to somehow get the best results of both situations. I don't think they have any method that can do that, though. You would have to know the intent of the user to know when you can go for data accurate to the training set versus data that is more accurate to possibilities. (For a fun example, Planet of the Apes have a Lincoln Memorial that I would imagine would be very hard to get from Gemini, as is.)