Hey there, time traveller!
This article was published 07/12/2022 (1435 days ago), so information in it may no longer be current.

Scrolling through social media over the past few days has felt like wandering an uncanny valley of friends and acquaintances’ faces. Faces I know, rearranged to look like anime or sci-fi characters or painterly works of art.

These images have been generated by the photo-editing app Lensa, which has gone absolutely viral after adding a “magic avatar” feature. The feature uses AI to render your photos and selfies into fantastical images and, for a few dollars, you can get 50 portraits of you. You as a fairy. You in space. (A lot of you seem to want to be astronauts.)

The appeal of this is understandable. People like personalization and instant gratification. It’s fun to see, via filter or app, what you’d look like as, say, a Simpsons character or a Disney princess. These images become someone’s profile photo on social media for a bit, and then we move on. These new images are like the apps of yore on steroids; the likeness of some of them is almost unsettling.

But I don’t know, guys. Are we maybe trusting the robots too much? Haven’t we learned anything from movies?

I’m not here to finger-wag, but “relax, it’s just a bit of fun” has a way of taking out any meaningful discussion about the implications of what this means for artists, or the ethics around it. After all, the images that AI is trained on come from somewhere, namely the work of real artists and photographers — and now, your literal face.

Many artists have expressed criticism that so-called AI art not only rips them off but devalues the work they do, and I tend to agree. It establishes the belief that art should be cheap and fast, when it is neither of those things.

There are a lot of artists, designers and illustrators in the world who cannot make a living wage, and whose time, effort and mastery is not considered when people balk at the price of a commission.

Hiring an artist to render me, say, in comic-book style for a cool portrait I can use on my socials or hang on my wall is obviously going to cost more than an AI-generated image on an app.

In June, Cosmopolitan unveiled the world’s first AI-generated magazine cover using OpenAI’s DALL-E 2, the popular text-to-image AI image generator. “And it only took 20 seconds to make,” chirps the cover copy.

Of course, that’s not entirely true — a team of women oversaw the process. But what happens when the AI gets better and faster? Whose jobs are on the line?

Play this out. What’s next, AI-generated album covers and book jackets? Followed by AI-generated songs and novels?

And perhaps this is cynical, but who will be inspired to even pick up a paintbrush or mess around on a tablet — and then spend the thousands of hours of work it takes to master something, while keenly feeling the frustration of being bad at something you desperately want to be good at — if a bot can just… do it for us?

Yes, tools such as DALL-E 2 certainly make it easier to extract things out of our human imaginations and get them down onto the page or screen, fully formed, almost instantly. The sample text descriptions on OpenAI’s website read like Mad Libs.

You want “an astronaut riding a horse in the style of Andy Warhol”? No problem. “Teddy bears working on new AI research as kids’ crayon art”? Here’s 10 versions of that. It’s creativity meets instant gratification.

But art is not something to be optimized or automated. Art is how we express ourselves, how we relate to each other, how we comment on and document the times in which we live — and that includes the act of actually making it.

If art is a foundation of humanity, then it stands to reason that humanity is a foundation of art — in any form, high or low. We’ve all seen the viral “I forced a bot to watch 1,000 hours of x and then asked it to write its own… ” memes. It surprises me not at all that the funniest ones are actually written by humans satirizing AI.

Largely removing humans from the process is likely why AI-generated images always look and feel a bit… off. Soulless. That’s why I call it “AI-generated imagery.” It’s not art, not really.

My current Facebook profile photo is an illustration of me, though it wasn’t created by AI or an app. It was gifted, not commissioned. The artist is my niece, then age four or five. Medium: pencil and marker on paper. You can tell the figure is supposed to be me because, while she only has a lopsided smile and empty eye sockets (haunting), she has a topknot. It is hilarious. And it is art.

I doubt the prompt “woman with topknot as kids’ marker drawing” would yield so charming — or so human — a result.

jen.zoratti@winnipegfreepress.com