AI was already used to determine sentencing requirements and recommendation and at that, those scientists basically ignored that racism played a large role in its recommendations.

None of this bias stuff is new and it's a basic fact of training an algorithm.

Just like budgets are moral documents, training is a moral system