If you want moderation, you need human moderators.
The trouble with the tech giants is that their "free service paid by ads" business model completely collapses into red numbers if they start employing adequate numbers of people for that purpose.
So they resort to artificial intelligence, which is worse than natural stupidity in this regard.
Not only that human moderators are better than ML moderators, but we probably want human moderators enmeshed in the community. Having Filipinos moderate the speech of Arizonians is not going to work very well.
There should be a certain degree of cultural competency, though.
For example, Czech Facebook banned an ad with a word "Rifle" in it, because of an American ban on gun advertisements. But "Rifle" means "Blue Jeans" in Czech and nothing else. It is not a gun-related word in our language. And indeed the ad tried to sell blue jeans.
The trouble with the tech giants is that their "free service paid by ads" business model completely collapses into red numbers if they start employing adequate numbers of people for that purpose.
So they resort to artificial intelligence, which is worse than natural stupidity in this regard.