Facebook's AI Can Analyze Memes, but Can It Understand Them?

When most adult humans are confused about many of the n-meta memes floating around, I can only assume 'no'. Perhaps there may be some interesting insight gained, but ai still can't 'understand' perfect speech, much less towering levels of irony and inate human reaction.

I suspect there are humans grading each one for 'hate' or whatever, and then letting the "ai" find variants and flag them automatically afterward.

It all reminds me of those (hopefully tounge-in-cheek) papers analyzing 4chan, SA, or other forums. I miss pre-clickbait wired.