How to tell a distraction from an endeavor
I’ve been writing and illustrating flipbooks. My favorite one is Skirting — go read that if you haven’t already. Now that I’m done with skirts, I’m working on fruit. But I’m constantly worried that this is a just distraction from doing something I’m already good at.
I skipped so many traditional skills by going straight to drawing on my phone. Or painting. Or whatever you call it when it’s your finger. At this point, I can make realistic fruit. I’m basically a 1960s throwback, channeling Gerhard Richter. He’d project old photos onto walls and trace them large enough to perfect every detail in oil paint.
I’m using his technique but on an iPhone. For a few weeks I amazed myself. Then I thought: who cares? AI can do this.
But I remembered that Richter started smudging his paintings – maybe because that was the only way you could tell they were real.
He also grew up in Nazi Germany, so smudged is probably how he sees the world. Regardless, I love the idea that ruining something makes it real.
So I tried to squash my fruit. First I asked AI to do it. We have to make that our first step with everything now – it’s too absurd to do something AI already does well.
Amazingly, I’m better at squashed fruit than AI. But not by much. There are some things AI doesn’t understand. Like, AI is bad at creating details without patterns. I noticed humans have this problem too. Even as late as the 1790s, American painters were doing a lousy job with hair.
Also, you know how AI can’t do hands because it loses track of where the fingers are? AI has the same problem with cherry stems. I hadn’t seen stems and fingers as the same thing, but now I get it. It’s hard to keep track of appendages growing out of their spot but showing up somewhere else entirely.
So I’m working on squashing stuff. Maybe that’s what we’re always doing – picking and pulling at things to show ourselves they’re real.
Great take, P! A most useful reframe in trying times. I hide jokes in my works that only one or two souls will ever get. It’s like signing a canvas before you paint. When future AI claims it’s its, a UV scan will prove it wrong. A skirmish won as the war descends. Thanks for keeping us in the fray. P.
AI also can’t do a simple tessellation. It’s hilarious, because AI can perceive that it fails at it, but still can’t succeed at it. You could explain step-by-step how to make a tessellation and it still couldn’t do it. The mathematical reasoning required is impossible in AI because all they do is imitate, or average things out, not reason.
I had to google tesselation. But once I saw it, my first thought was: that seems built for AI. It’s very surprising what AI can’t do. The experience of doing a lot with AI is so similar to having a valuable co-worker. When someone at work is valuable they are never valuable for everything. But it’s our instinct to try to get them to do more and more and more that we are not good at. So many days I find myself talking to AI like it’s a recent college grad who I’m managing: “Are you stuck? Do you need me to explain it a different way?”