We've seen that bots can be trained in different ways. I could see trusting a bot to have a voice in how content is moderated, and then have humans to spot check the bots and deal with problems that are escalated. Bots are really the only way to do this at the scale that will be needed.