Fight Biases, or Route Around Them? — AI Alignment Forum