That is the point, to show how AI image generators easily fail to produce something that rarely occurs out there in reality (i.e. is absent from training data), even though intuitively (from the viewpoint of human intelligence) it seems like it should be trivial to portray.
If someone ask for a glass of water you don’t fill it all the way to the edge. This is way overfull compared to what you’re supposed to serve.
Oh man…
That is the point, to show how AI image generators easily fail to produce something that rarely occurs out there in reality (i.e. is absent from training data), even though intuitively (from the viewpoint of human intelligence) it seems like it should be trivial to portray.
Omg are you an llm?