My pet theory is that both Google and Facebook need to create training set data for their proprietary AI content moderation algos. The content moderation algor

My pet theory is that both Google and Facebook need to create training set data for their proprietary AI content moderation algos. The content moderation algorithm is optimizing for three variables simultaneously - maximum number of videos hosted, maximum number of views, minimum number of lawsuits/negative press articles written/anti-Big Tech tweets. The algorithm can't know if its truly maximizing the first two without letting the last one trigger every so often. We are just witnessing/participating in the live dataset they are testing on.