QAPR 4: Inductive biases — AI Alignment Forum