This line of thinking leads to non-solutions. There are potentially many options beyond the dichotomy of crude AI automation and an army of human employees. Fi

This line of thinking leads to non-solutions. There are potentially many options beyond the dichotomy of crude AI automation and an army of human employees. First, I think most of us would just like a relatively transparent method of escalating issues that will eventually get to a qualified human. Put some time delays and recaptchas to stop spammers from abusing it but if someone is willing to put in the time and effort to fight for a video, they should get a clear response as to why it was taken down and who made that determination. Then there is the immense potential for crowdsourcing moderation. You can both get data by polling users (again with basic checks to prevent exploits) and promote certain users to a moderator role such that they can keep the community in which they participate reasonable. Especially for disinformation, how hard would it be to find volunteers to tell other people they're wrong on the internet? And finally, if erroneous moderation has the potential to cause many problems, it is worth re-evaluating how much is actually necessary. For legal reasons there is some content that you really need an AI to quickly scan for, such as child porn, but in a lot of cases the harm in leaving up a video is insignificant. There are a lot of sites on a similar scale to youtube that have substantially better (even if not necessarily perfect) content moderation despite working with substantially lower resources. If they can do it, so can Google.