No I was asking about how do would they measure the bias of the training data in the model. You responded with "hurr durr bias testing bad because of one incid

No I was asking about how do would they measure the bias of the training data in the model. You responded with "hurr durr bias testing bad because of one incident." Now, I am 99.99% sure that you think that the "mainstream" news sources are liberal-biased. All of this news text and transcripts were used in training these models. Wouldn't it be in your interest to measure how much it affects the output?