The Innocent A-Eye
AI generated images are creating an environment where we can't trust what we see
“The innocent eye is a myth.”
-E. H. Gombrich, Art and Illusion
The other day, I posted a photo to my Substack that I’d found on Flickr Commons. I labeled the photo roughly as I found it in the Commons, “Fulton Street Dock, Manhattan skyline, Manhattan – 1935.”
18.5 hours later, I got a comment from someone that said: “Is that why there is a fat dude in a hoodie, a guy on an electric scooter from [sic], and an ai generated half sunk ship?”
Huh. Now that you mention it, I see it, too.
The fat guy in the hoodie is in red, the guy on the scooter in yellow, and the “ai generated half sunk ship” in orange:
Ignore the ship. The picture was taken in a harbor, so a ship wouldn’t be out of the ordinary. For now, just focus on the figures outlined in red and yellow.
Indeed, these figures do look as described. Granted, they’re a bit blurry. But the guy in the background does look like he’s on an electric scooter, and the chap in the foreground looks like he could be carrying a phone.
The strange thing is, I know this photo is legit. It came from the New York Public Library. See for yourself:
Upon closer inspection, you can begin to parse out what’s really going on. The chap in the foreground is in galoshes. If you look over the shoulder of the scooter guy, you’ll notice someone else pushing a cart or dolly.
So, either one of two things is happening here:
The New York Public Library uploaded an AI-generated image, and the photographer Berenice Abbott (who took the photo) doesn't exist.
We're seeing things that aren't there, and AI image generators are changing the human eye at a fundamental level.