AI having biases (or just weird opinions) isn't necessarily a big deal in itself, the problem arises when we start delegating important tasks to AI such that th

AI having biases (or just weird opinions) isn't necessarily a big deal in itself, the problem arises when we start delegating important tasks to AI such that those biases begin to have real, harmful consequences. I agree we probably shouldn't expect AI to have the "correct" moral opinions, but then we also shouldn't allow AI to determine what content gets censored or promoted, who gets punished, what job applications get prioritised, etc.