Why I’m Done Using And Boosting AI Art

From Chuck Wendig: Terribleminds:

Let’s just put it out there and up front — earlier, I was glad to play around with AI art, but that has ended. I have no intention at present of mucking around with AI art, signal-boosting it, or supporting it. I had a subscription to Midjourney, and I canceled it.

Now, to rewind a little —

I think AI art is pretty cool.

I know, I know — I just said, but I won’t support it, and that’s true.

But I think it’s neat, in a general sense. It’s like, we can make COMPUTER ROBOT GHOSTS do all kinds of cool things for us — they can tell us the weather, show us how to get to the mall, I can yell at my car to turn on the heat and it’ll totally do it, Gmail can already predict the response I’m going to make and start to prep it for me. The robot ghosts are cool. So, the ability to say, HEY ROBOT GHOST, SHOW ME WEREWOLF PIKACHU USING A NEW POKEMON MOVE CALLED “CORUSCATING ELECTRIC ANUS” ON A KAIJU VERSION OF JERRY SEINFELD and then somehow it sorta does it, well, I don’t hate that.

Now, admittedly, when I started mucking about with AI art in the long-times-ago epoch of, mmm, six months ago, what it produced was often fiddly and hilarious and straight-up ****** weird. It would still have eyeballs in places where there shouldn’t be. Some guy’s face might look like a smear of paint, and his hand would have sixteen fingers. You might squint and see Sophia from the Golden Girls mysteriously hiding in the wallpaper. It felt a bit like you watching a robot dream. Like you were privy to the growth of its creative mind.

(It’s a lie, of course. There’s no robot dreaming; that is a romantic, anthropomorphic notion.)

But it didn’t take long for the results to get… good. Real good. Freaky good. You plug in something and what returns is a foursquare array of nearly exactly what you asked for, in a variety of art styles and modes. Which, one might argue, is quite the point of this whole affair, and I suppose it is, though I’ll also note for my mileage it also kinda defeats if not the point, than rather, the delight of having a robot puke up something just super ****** weird instead of precisely what you asked for. We were training the robot well. And it was learning fast.

And now, you see the so-called AI art everywhere, and you also see those who are mad at so-called AI art everywhere. And the latter category is often artists. Not always! But often enough.

As such, I’m going to side with the artists.

Link to the rest at Chuck Wendig: Terribleminds

PG wants writers and artists to live long and prosper and thinks that they will do just that. While he understands how they may worry about AI tools diminishing their income, he doesn’t think AI will replace them.

4 thoughts on “Why I’m Done Using And Boosting AI Art”

  1. Faux AI art *is* going to displace a whole category of painters. The philipino(?) “starving artists” producing high volume originals.
    That much is unavoidable.
    On the other end, high talent creatives will endure as they always have.

    The commercial artists in the middle? Not too sanguine about their prospects.

    These tools are coming at a time of economic pressure that increases their value.

    For example, a writer friend of mine pointed me at a web-based operation selling at good prices original pre-made art meant for genre covers and doing commisions for £500. It doesn’t say where the art comes from, a boiler room or software. As a business, there is more profit (and likely better prices) in the latter.

    It may be disquieting but the trends are there; a skilled writer of prompts can easily crank out a dozen or so covers a day, sell them at $100, and clear a cool half million a year all by himself.

    The OP might feel good about his announcement but an Indie author starting out may not be able to afford such qualms.

  2. We need something like a Turing test here. Make a group of 100 images, some created by humans, and some created by AI. Then subjects tell us how each image was created.

  3. There are all sorts of places for commercial art. I don’t look at AI as replacing covers, in general (though I’m sure it can, I value and am willing to pay for the uniqueness and recognizable consistency of a particular artist for my Fantasy genre series entries). But I think AI greatly liberates the minor uses of commercial art for writers, namely advertising images, blog images, and so forth.

    • Like so many things, the proper answer is “it depends”.
      Having a familiar go-to artist to provide a consistent trade dress to a fantasy series is valuable.

      But for standalone books in most any genre?
      Or how about romance, where different subgenres have common cover layouts? (Young woman in 18th century formal attire standing on a hill across from a ridiculously good looking young man in a kilt?) 😉

      Or how about the stylized covers in contemporary light romcoms or cozy misteries, like these:


      eBooks removed the single biggest barrier to entry for newcomer authors (agents) but covers and editing often remain a stumbling block. We have yet to see a genre aware proofreading “AI” tool but I have little doubt one is on the way. Not a spell checker or grammar checker but a competent beta reader that can flag loose threads, inconsistensies, and contradictions, tunable for tone and style. A matter of time.

      But covers, that looks to be close to a solved problem, going by the current state of the (ahem) art. And to a newcomer writer hoping to sell a few hundred ebooks in their first year, saving four or five hundred dollars and still getting a marketable cover *to their taste* via an “AI” cover is going to be a godsend in many genres.

      Not for everybody…
      …but something for a lot of somebodies.
      Which is what the OP fears. But that djinn is out of the bottle.

      (BTW, I just ran into a much more obnoxious captcha.)

Comments are closed.