Animated frames are supposed to convey intention. They’re fantastic at doing this since you can manipulate every detail of every frame. The idea that you’ll just run an AI through it, that might work for dialogue scenes of a typical Japanese TV anime where intention is low and mostly it’s indeed grunt work. But I would imagine it would be a bit lifeless - unless someone trains an ML specifically for anime using good animation as a reference.
Basically just moving between two frames is an example of extremely poor animation.
Source: am animator, sort of.
In an animation those two photos would be drawn and created as keyframes which would then get interpolated many ways (hopefully not linear and as robotic and weird as this).
Very interesting technology though. I could see this coming to an smartphone near you any day now. And there will be ways people animate with these tools but this isn't it.