To me the intrinsic problem with slop is simply quality. In areas were (I think) I am qualified to discern high from low quality (because experience) I am not impressed with the output of the bots, at all. So I would be fooling myself to think that in areas where I am not qualified, the bots actually do a good job of producing an end result. That's why I don't use it to produce any output for me, but instead, apply an intermediate-output-only policy.