Source: James Bridle at medium.com
On the surface this sounds like “Won’t somebody think of the children!” type alarmism:
What I find somewhat disturbing about the proliferation of even (relatively) normal kids videos is the impossibility of determining the degree of automation which is at work here; how to parse out the gap between human and machine.
But when you look closer it really seems like there is something very creepy going on. Consider this description of one of Bridle’s examples:
The video consists of a regular version of the Finger Family song played over an animation of character heads and bodies from Disney’s Aladdin swapping and intersecting. Again, this is weird but frankly no more than the Surprise Egg videos or anything else kids watch. I get how innocent it is. The offness creeps in with the appearance of a non-Aladdin character —Agnes, the little girl from Despicable Me. Agnes is the arbiter of the scene: when the heads don’t match up, she cries, when they do, she cheers.
It’s a little off, but off nevertheless. This feeling seems to increase with each example, until the final one is downright nightmarish. Not nightmarish in the sense of, say, Stephen King’s It, but in a more literal sense. Like a scary dream with the accompanying twisted logic.
I think what worries me most here is this: These video’s are designed to apeal to children, and it’s easy for us to see as adults why this would work. Again, as adults it’s easy to see that there’s something very twisted and wrong here. In both cases: easy for adults to see this in videos aimed at children. But what happens when the techniques and algorithms1 get good enough to target adults? Will that be as easy to perceive?
- I’m assuming there’s some sort of machine learning involved in the creation of the videos. ↩