Nathan Lambert believes that attempting to regulate AI model distillation through government policy would cause more harm than good