Yeah we know positive reinforcement is better than negative one for humans, why wouldn't you use the same approach with LLMs. Also it's better for your own conscience.