The State of the Art

Why A.I. Caricature is a Bad Replicant of the Real Thing.

This week on social media, I had a discussion with an A.I. enthusiast about the merits of A.I.-generated caricature. My position was simple: there aren’t any.

The enthusiast posted a string of software-generated celebrity caricatures, and waxed lyrical about how good they were. But they weren’t. They were terrible. As I pointed out, while some of them bore a vague resemblance to the person they were supposed to portray, that's still a million miles away from making them fully resolved and successful caricatures.

He then went on to suggest to me that the average Joe’s ability to have ‘just a bit of fun’ with software-generated caricatures actually drives interest in acquiring commissioned pieces from real-life artists. But this is a false syllogism. It's tough enough persuading people to buy real art. When A.I. doles out prompt-generated freebies it only serves to dilute people’s already weak willingness to approach artists, making the marketplace an even tougher environment.

To me, A.I. caricature is symptomatic of a wider obsession with turning the creation of art into a quick fix. But great art comes from protracted effort, trial, and error. When someone asks me how long it took to create one of the caricature sketches I post online, I usually tell them an hour and thirty years.

But I have to be honest and admit that we caricaturists must shoulder a big chunk of the blame for the generally dismissive, ‘just a bit of fun’ attitude towards caricature. It pains me to say it about my chosen art-form, but the majority of the human-generated caricature from which A.I. learns to create its likenesses is, frankly, terrible; ergo A.I. caricature is also terrible.

The problem is that Joe Public usually doesn't know - or care to know - when he's looking at bad caricature, however it's made. And, sadly, the same can be said of many caricaturists.

I feel that standards in this art-form have been in decline for many years. And caricaturists simply aren't good enough at evangelising for great caricature. I'm doing my part to try to help change this. My hope is that other caricaturists will join me in the discussion about improving the current state of the art.