Generative AI, Which Is The Future Of Art, Cannot Draw A Map Of The United States

It truly must be seen to be believed.

1:35 PM EDT on July 4, 2024

Maybe you’ve been on Twitter and seen the video of “stunning AI spaghetti art” that one gormless loser has claimed is proof that generative AI will one day win Oscars and Grammys. This is what I do every time I hear that claim: I ask ChatGPT to draw a map of the United States.

Before I show you the image that ChatpGPT produced for me today after being given the prompt “use ascii art to draw a map of the United States,” I want you to imagine the contours of the United States in your mind. It’s July 4th, after all, a day when we celebrate America’s freedoms, such as … and … and who could forget …. You might not be able to remember all of the states names, but I can picture the Florida panhandle and the weird triangle part of Texas and the curvature of California. Hold that image in your head, and now scroll down.

Ascii art that's a series of rectangles with some state abbreviations like ME and VT scattered throughout

Is this a fluke? A one off? No. Here’s another response to this prompt that looks like a little dog.

Ascii art that looks like a dog wearing a dress with the caption: This representation captures the general outline of the continental United States, including some key details like the East Coast, Gulf of Mexico, and the basic shape of states like California, Texas, and Florida. Each part of the map is approximate and simplified for the ASCII format.

Generative AI is stupid. It is not a thinking machine. Art is made by people who can see and interpret symbols, and then re-interpret them into new images. At the very least it is made by people who can picture the United States in their mind and draw a vague outline of it when prompted. All the rest is stunning AI spaghetti art.

Enjoyed this article? Consider sharing it! New visitors get a few free articles before hitting the paywall, and your shares help more people discover Aftermath.