DALL-E doesn’t understand the absurd. So, when it gets a prompt that connects things that did not, could not exist together, it will exaggerate the absurd — most of the time.
Also, acient Egyptian art always fascinated me. Not that I’m such an expert. But it really takes the viewer to a world where people saw things completely differently.
I simply had to try this:
Me: Create an old Egyptian wall painting that shows a group of experts arguing about the risks of creating artificial intelligence.
See what I did there? Also, see what DALL-E 3 did there? There is not an ounce of “artificial intelligence” in the image, but it also does not look like an Egyptian wall painting. Especially the first image: it looks like a crossover of a photo and a painting of people in old Egyptian attire, in front of a painted Egyptian wall. Does this have to do with the modern reference introduced by my including AI in the prompt?
I didn’t take the pain of making an experiment, but I remembered that I had made a similar attempt with old DALL-E 2. This is what I got:
Me: An old Egyptian painting of a panel of experts thanking their audience.
(You did not do dialog with DALL-E 2, just threw in descriptions.)
Finally, you can always mix it up by adding cats:
Me: An old Egyptian painting of a panel of experts thanking their audience of cats.
Well, DALL-E, who are the cats, the experts or the audience?
thirdtower.ai 