Does any quasi-xml work, or do you need to know specific commands? I'm not sure how to use the knowledge from this article to get chatgpt to output pictures of people in underwear for instance.

It looks like later in the article they drop some of the pseudo-xml and it still works.

I wonder if it’s something like: the model’s training set included examples of programs configured using xml, so it’s more likely to treat xml input that way.