kinda amaze me that bias is a neutral feature in neural networks, because otherwise it's pretty hard to tell people that all thoughts are biased, just that some biases are currently unacceptable.