The days of needing third party moderators should be coming to a close. 

I see no reason why it shouldn't be possible to train an AI agent on what types of content and replies you prefer not to see, and hand off the heavy lifting to the AI. And if the AI screws up, you simply give it more training!