President Trump’s explosive revelations about creating a Riviera of the Middle East on redeveloped Gaza land seemed ripe for an old-fashioned political cartoon.
What happened next is a bit revealing about how ChatGPT (and maybe other GenAI tools) use prompts to create an end-product.
After President Trump spoke on Tuesday from the White House, I asked ChatGPT on my phone to create an illustration of a man looking like Trump wearing a golden robe and standing at a balcony amid highrises overlooking the sea.
After a few seconds, ChatGPT responded that the request violated its rules about public figures. The explanation was lengthy.
At that point, I thought the concept for what might become a political cartoon was a dead end, but then ChatGPT surprised me and said it could create the image I wanted with an older man standing at a balcony.
Of course, I said to go ahead. ChatGPT came up with an image similar to the one you see above. I was hoping to create the image as a futuristic look two decades ahead, when Gaza could be recreated under the grand Trump vision of a Rivera of the Middle East.But what ChatGPT gave me back was an image with very futuristic, sci-fi highrises. Then, I asked for less futuristic highrises and ChatGPT came back with the actual image above.
After that, I asked for an older man without glasses, but what came back was the following:
That new image seemed too much of a commentary on the old man’s expression and why was he looking so dyspeptic, but also: why was he facing us?
It felt to me that ChatGPT in this latest iteration was really taking on the role of editorializing more than I expected, with the central figure confronting us, full of questions and qualms and challenges. I didn’t really intend to go that far, so I shared on a couple of my personal social accounts the image of the older man with glasses looking out to sea. The cutline was, “Gaza Riviera circa 2045 as seen from the Trump Tower penthouse.”
In sum: Some comments by political figures are just too amazing to be ignored and that’s what journalists are born to confront. There are also some interesting questions of how chatbots within the Generative AI realm enter into a dialog with humans over content. By now, with such tools in fairly wide use, there must be plenty of similar examples that all of you are aware of.
I offer this modest experiment to show what’s possible, for free today, especially if you or your co-workers aren’t as plugged in about the capabilities of chatbots, GenAI, agentic AI and what’s coming to your doorstep with AGI.