Here’s why I’m just saying no.

I’ve been playing with AI lately – specifically MidJourney, a “generative artificial intelligence program” that creates art* using prompts supplied by the user.

I started for innocent enough reasons. If you follow me on Facebook or Instagram, you know that most days I post a Zen-oriented meme which features some brief text – a haiku, perhaps, or a selection from my book – along with an image that complements the thought. Like this one:

The words are easy enough, but the images were often a problem. I wanted something that suited the text, and occasionally I found what I wanted online. Often, though, the images that worked were copyrighted or owned by a stock warehouse. And I didn’t want to steal.

So I thought that being able to create unique images would solve my problem. And it has.

Thing is, these generative AIs are a lot of fun.

It’s absolutely insane what you can do, and so I started experimenting. I learned some basic prompts and rules, and over the span of a month I cranked out some pretty cool images. I didn’t tell myself they were art, because I didn’t really do anything except figure out how to say some words to the machine. And as someone who’s fancied himself an artist since the late 1970s, I have some principles that matter to me a great deal.

Then I discovered the part I didn’t know at first: not only will MidJourney work with text prompts, it will also take reference images. That is, you can give it a picture as well as words, and it will build that into the output.

So I started giving it some of my photos to consider. Most of it was forgettable, but a few were absolute grabbers. (I thought, anyway, and a lot of friends agreed.) Like these:

I began to think about justifications for this new art style.

Aside from the usual (AI is just a new tool and there have been zillions of other disruptive new tools in the past, anything you do will be clearly labeled, so the audience can decide, etc.) I decided that these were different because they began with my art (I used my poetry in places, too). The result, therefore, was a true collaboration between the machine and me. It could not produce any of this on its own.

Interesting idea. Seductive idea. And I went so far as to begin planning a new web site that would be devoted to this work.

And now I’m stepping away.

Two reasons. First, I never stopped thinking about the artistic, critical, and ethical concerns surrounding artificial intelligence taking on work previously done by humans. It was always on my mind. Then I encountered a column in the LAT:

Your boss wants AI to replace you. The writers’ strike shows how to fight back

I was especially struck by this quote from artist and activist Molly Crabapple:

“I saw my work in the LAION-5B dataset used to train Stable Diffusion,” Crabapple says. “I saw DALL-E’s ability to churn out bastard versions of my work with the prompt ‘drawn by Molly Crabapple.’ I saw how tech corporations, backed by billions of dollars, had gobbled up my work and the work of countless other artists to train products whose goal is to replace us.”
…
“There is no ethical way to use the major AI image generators,” Crabapple says. “All of them are trained on stolen images, and all of them are built for the purpose of deskilling, disempowering and replacing real, human artists.”

For the sheer hell of it, here’s MidJourney’s take on [Molly Crabapple drawn by Molly Crabapple].

The article (which is well worth a read) and Crabapple’s comments brought me back around to the debate I was already having in my head and to my thinking about diving into this new AI art thing “seriously.” Also, it led me to my second reason for stepping away from the machine.

Let me illustrate this one. First, here’s an AI piece I call Phalanx:

Not gonna lie: I like it a lot. Now, here’s the original photo the AI was working with – my contribution and my justification for why Phalanx might, just maybe, be defensible as art. It’s called Lift Up Your Voice:

This is one of my favorite photographs, and others have said they’ll like it, as well. I’ll leave you to your own judgment, but when push comes to shove Phalanx just doesn’t look much like the photo, does it? If I tell you that it is a source for the AI image, you can probably look closely and see…some connection?

But whatever I tell myself, the AI version is one part me and 50 parts someone else. Someone like Molly Crabapple, perhaps. And if you look at the reference images I contributed for the pics above, you’ll find the same thing.

Now what?

[CAVEAT: A friend who read this draft said my tone seemed a little self-righteous, and he may be right. That isn’t the intent – in truth, I’m a little disappointed in myself for falling prey to the charms of the digital temptress. But, as always, YMMV.]

Yeah, I’ll keep using MJ for my daily Zen reflections, and since it’s fascinating, I’ll probably keep playing with the AI as new versions are released.

But no web site, no winterSmith+AI “art” for sale. Essentially, I’ll never use AI in a way that has the potential to take money out of the pocket of a fellow human being. If I hit the lottery, maybe I’ll hire real illustrators to help me with my Zen meditations, even, but for now there’s no money in it for anyone.

But others are out to fuck you for fun and profit. Mainly profit.

Like I say, read that article.***

______

* I footnote “art” not to start an argument, but because whether or not its output will attain that critical status is far from determined. It will take years, maybe decades, maybe longer, to sort that one out.

** I know, it isn’t real artificial intelligence. Let’s save that for another day.

*** I’m a business writer for a living. Yes, I have heard of ChatGPT. And yes, I’m paying attention.