...but newer, larger multimodal LLMs appear to be on track to understand the photo, to find it funny, and to be able to describe why it's funny.

We live in interesting times.