I can't really say you're wrong because AI advances really quickly, and with the improvements in AI-generated art, I would lean that you're at least not far off.

The challenge with understanding is that while it might just be photo + text, take some of the LOTR memes, you can add text to almost any scene, the text might relate to the image, the scene the image is drawn from or drawn on the emotional sentiment of the scene, etc.

So, there would be plenty of opportunity for error in understanding because there are layers of understanding that might need to have an understanding of the film.

Producing memes would be much easier from that perspective.