Assembly is the most imperative language. Are AI prompts the most declarative?
This is a thought experiment, not necessarily my opinion. The definition of "most" is intentionally left ambiguous. Please be respectful.
That makes them hard to compare.
So far AI image generators don't have a lot of understanding of syntax. They struggle with something like "four red cars above five blue cars". Rather if you prompt them with "darts vader drinking beer in the backrooms" the system activates "darts", "vader", "beer", and "backrooms" in different parts of the image with certain kinds of consistency checking involved in selecting images to show you. Most of the time people like the results but it's not guaranteed to do anything at all other than consume resources and make some kind of image. Contrast that to an assembly program which could be proven to do some task, say sort a list, correctly.