Is AI Art Bad for Artists and Creators?

A robot in love
A robot in love

I hated my university’s slogan. It was something along the lines of “Thinkers. Leaders. Doers.” It just rubbed my brain’s edges in all the wrong ways.

Doers.

Blegh. I’ve never been a doer. I’ve definitely been a thinker, though. Some might even call it chronic. “What are you thinking about?” is a phrase I’ll hear at least thrice a week. Sometimes even when I don’t leave my cave. Although those moments have the mirror worrying a bit more than usual.

On a recent trip with friends, I confessed I had been thinking about where to park before we even left 2 hours prior.

“I don’t think like that,” was the reply.

“Hmm. I do. It’s the only way I can.”

But if I had to say I’m good at something, it’s conceptualizing. Organizing. And never getting around to fucking doing it. If someone gave me $1 million to make “Procrastination” my middle name, I’d chase after a blue car before they finished the question.

I mean hell, did you see the cool white stripes on it?

Then last week I listened to an episode of one of my favorite podcasts, You Are Not So Smart, and went down yet another fun rabbit hole.

The guest was a wonderfully accomplished woman named Temple Grandin, whose list of feats is no less than incredible:

  • She has a degree in psychology and a PhD in animal science.
  • Authored/Co-authored over 60 academic papers.
  • Time 100 named her as one of the 100 most influential people in the world under the “heroes” category.
  • She never spoke a single word until the age of 4. At 2, she was (falsely) diagnosed with brain damage.
  • She was later identified as an autistic savant and has been a successful advocate for autism and neurodiversity rights.
  • She even had a movie made about her life.

She truly sounds like an incredible person.

But the part of that episode that intrigued me was her theories about how people’s brains work. Her idea is that some people intuitively think visually, rather than using words to form thoughts and ideas.

As in, when reading a sentence or listening to someone speak, the first thought in a person’s mind is a picture rather than words. Honestly, it struck a bit of a chord with me.

She later went on to discuss her theories that visual thinkers are probably more prone to dyslexia and other common problems, not because of a lack of intelligence, but in how we are taught in our modern education system.

How People Think With Words or Visuals

Learning to read and interpret little figures on a page and absorbing them into complex patterns of ideas is an extraordinary feat. I’ve studied Mandarin Chinese on and off for the better part of my adult life, and trust me, though I never had dyslexia as a kid, you should see me try to read “讀者,你真的去翻譯了嗎?”

And to the extent it’s been studied, the science apparently backs her up, at least in the form of how we think.

According to child development theorist Linda Kreger Silverman, her studies suggest:

  • 30% of the population relies on visual/spatial thinking
  • 45% use both words and visual/spatial thinking
  • 25% use only words

Some other research (which I can’t locate right now, sorry!) suggested the visual/spatial thinking population has a tiny extreme, where people are almost unable to think in any other way.

I don’t know where I lie on this scale, but one specific memory does stand out from grade school. I had just come home from class after a big math test (aced it, of course, woo!) and was telling my very busy mother how I could see the numbers fitting together in my head as kind of a 3D puzzle.

It wasn’t the usual “7 + 8 = 15.”

To me, the top part of a chunk of 7 fit neatly together with the leftover part of 8 to form a “block” of 15.

I always equated it to something like Tetris. Not really sure how to explain it better than that. Anyway, she advised me to stop daydreaming in class.

Made sense to me, but I didn’t stop. I couldn’t. It’s just how my brain wrapped around complex stuff. I think it’s why I used to be a good product manager, but a dry sponge when it came to presenting my plans — I soaked up all the energy in the room.

Luckily I had a boss that “got it” in private and advocated for the projects on my behalf. He was an engineer.

And according to this conference paper presented at the Technological University Dublin:

“In tests with more than 30,000 professionals, engineers demonstrated the highest level of spatial visualization skills, followed closely by architects and other STEM professionals. A recent study shows a strong correlation between spatial visualization skills and creativity and technical innovation.”

And the dominos eventually fall into place.

Artificial Intelligence

Not the true AI, of course. But businesses being the slimy assholes they are, of course they gave a new fancy technology a fancy name it doesn’t really deserve.

I’d go into the history of it but I’ve probably already typed enough. I’m sure you know what I’m talking about. AI to write our poems. AI to draw our dreams. AI to program our Zoombas to make breakfast in bed.

I got interested in the writing one almost two years ago. I was excited to try the new improperly named software and spent almost 2 weeks sucked into its vortex of promise:

What can it do? What can’t it do? Can it do the parts of writing that I find extremely tedious? Can it help me do it faster? Will it keep my voice? Can I now create tutorials on how to make breakfast on your floor with a Zoomba but written by Shakespeare, even though I never particularly enjoyed his writing nor eating off the floor?

Needless to say, the next time daylight shone upon my foggy eyes, my face was full of disappointment.

It could not do those things. Not well anyway. Not yet.

But that’s how technological progress works. We can’t stop it. If there’s an incentive and some ADHD-riddled scientists working on it, it will eventually happen.

Then came DALL-E and MidJourney and Diffusion and my passion for visuals flared up like a bad case of floor E. coli food poisoning. I’ve always loved the idea of being able to draw, but never took the time to actually learn it. Maybe I just never had the right opportunity or motivator for it. I dunno.

But MidJourney, oh, she’s my baby.

With MidJourney’s help, “I’ve” created thousands of images. Sometimes I’ll sit down and get lost in her wonderful little world, and peek back out with a work of art that would’ve taken three days to create IF I was a full-on expert graphic designer.

Newsflash: I’m not.

Case in point #1: My 3rd painting during 3 months of trying to learn how to paint on the weekends:

A portrait of a friend

Case in point #2: After spending a year digitally drawing for fun.

Catcy Reagan

That took me over 10 hours. Yikes.

Needless to say, being an artist wasn’t written into my future.

MidJourney Artificial Intelligence Art Generator

But let’s jump back to Midjourney, the artificial intelligence drawing software.

For those unaware, most of these AI programs work by gathering billions of datapoints, then through machine learning, iteration, and an immense amount of effort + time, try to construct new content based on what it “thinks” we are asking it to do.

Ask MidJourney to draw me a puppy and after previously having consumed thousands of pictures with “puppy” as an associated text, it then constructs, pixel by pixel, what it thinks I mean by a puppy.

And then it gives you the options, oh boy, the options.

4 fluffy puppies

Each iteration is new, and more importantly, modifiable, at least to a degree. The process can go on and on into infinite, granted your shiny object needle is still pointing in the right direction.

And what you get is nothing less than breathtaking — usually.

It can create a magical puppy for me, if I ask it to.

4 magic puppies

But it cannot create a magical puppy sitting on a magical cat.

4 magical puppies

Just like all of us, it has its limitations. Understanding multiple objects is a challenge for the program, but she won’t tell us that. MidJourney is shy.

Or hop on over to the newest version of OpenAI’s text generator called GPT-3 DaVinci. It is nothing short of amazing.

Here, I’ll let it introduce itself:

Introduce yourself (My text prompt)

Hi, my name is Alex and I’m from California. I’m an outgoing person who loves to travel, meet new people and try out different experiences. In addition to that, I enjoy learning about different cultures and trying out new things.

Again, they still clearly have their limitations.

But ask it to write a promotional poem for my newsletter, and:

Pryor Thoughts, a newsletter of sorts,

Will provide you with thought-provoking support.

It’s full of wisdom and insight for sure,

To help you on your journey evermore.

From the ideas it shares each day,

You’ll find new paths to explore along the way.

Stay informed when big news breaks out;

With Pryor Thoughts you won’t be left in doubt.

Subscribe now and explore what life has to offer;

Get ready for some exciting knowledge uncover!

Damn. Not sure I could write it better myself. I didn’t write a single word in that text block other than explaining specifically what I wanted it to write for me (the prompt).

This process seems to repeat.

Text AI works on billions of words and sentences and associated output. The first generations absolutely suck. But they refine it through user testing, panel judgment, and who knows what other methodologies, but eventually the next generation comes out and shocks people once again.

There’s probably a limit to how good they’ll get, or at least in terms of how much better the next generation can get vs the time to get there. But even right now, you can plop over to any of these AI websites, sign up, and for a fraction of what this work might normally cost in both time and money, start creating your own visions.

Is AI Art Stealing?

This is the main point of many an artist and creator. I recently went down a rabbit hole with a friend to see if we could create a cartoon book from a finished draft we had lying around.

I popped back out after 10 or so hours with, well, something I’d say would be a perfect mockup for a real artist to create off of, but nothing close to a book I could release to the public.

But that didn’t stop this gentleman, however.

He claimed to have created the world’s first children’s book entirely from scratch with AI (MidJourney and ChatGPT) all in the course of one weekend. Believe me, the thought of making that claim did come up a few times with my writer friend in our experiment.

It was inevitable, not an “if” it can be done.

As for a guess on how it turned out, well, just take a look at most of the reviews that add up to 2.3/5 stars.

Artists were not happy.

Why would they be? The thought of a robot taking over your entire job is disconcerting, to put it mildly. Unless, of course, you own the robots or the factory using them.

Sell the shovel, don’t dig the gold, or some other capitalist saying.

But there is a valid point. Most of these AI programs were trained on billions of data points that were used without asking the creator’s permission. That’s the main point of contention I see pop up — that it’s outright stealing.

As this artist who worked on several award-winning indie video games told Kotaku:

“The endgame of a potential employer is not to make my job easier, it’s to replace me, or to reduce all my years spent honing my craft into a boring-ass machine learning pilot, where I’m trained to vaguely direct an equivalent software in hundreds of different directions until by chance it spits out an asset we could feasibly use in a game.”

But an often misunderstood point is that they’re using real images and modifying them, rather than creating from scratch. Then again, you can upload an image, even a copyrighted one, type in a prompt to modify it, and out pops something unique yet recognizable based on the original artist’s style.

That, in many senses of the word, is stealing.

And if you go ask the lawyer bloggers what they feel, well, after reading dozens of articles on the subject, they only have one coherent message — we have to wait and see.

In the US, it seems the companies using these massive databases (think of copyrighted places like Shutterstock or open-to-use places like WikiMedia Commons) as their initial data set are probably covered.

According to an article in Engadget: “Text and Data Mining…is the technical term for the training of an AI by plowing through a vast trove of source material looking for patterns…in the US, TDM is broadly covered by Fair Use.”

And the organization that takes care of creative commons (ever see an image say CC BY 3.0?), which strives to abide by applicable laws, feels somewhat the same.

But go tell any artist who isn’t famous or selling their work for thousands or millions of dollars on the above points, and you’ll be lucky if you don’t get stabbed in the face with an overpriced collectible pencil.

Believe me, I have a few half-time artist friends who make a substantial part of their not-so-huge incomes from authentic artwork. Most of them feel either a strong sense of anger or a feeling of hopelessness.

“That’s it, AI took my job,” one told me.

I feel otherwise.

Will AI Art Put People Out of Work?

I keep hearkening back to the past. “History doesn’t repeat itself, but it rhymes.” With the invention of photography, realist artists around the world feared their skill sets — often taking a lifetime to perfect — were suddenly rendered useless.

Were they put out of work? Some, I’m sure, were.

But others adapted. They invented new ways of painting that cameras simply couldn’t capture. Impressionism and modernism came about in some ways as a response to this.

New forms of expression were created when technology adapted to the old.

A “photographer” was now a paid thing. And models. And camera companies. And workers for the new factories. None of these things existed before, and the entire world was blessed with images we would’ve never seen before — both artwork and pictures.

Perhaps that sounds blunt, maybe it’s the economist in me, but with all forms of technological change come changes to our everyday lives.

New jobs will be created, old ones will become specialized, and some obsolete. A horse trainer in 1850 wasn’t a particularly high-paying job, but go buy your child a pony for Christmas and you’ll see how much it is today.

To that end, I think there’s only one feasible option: adapt and overcome.

Most of the writer and artist friends I mentioned are particularly adept at unusual skills. The one sells book covers as he’s been a long-time collector and learned how to create them on his own years ago. He can now make astoundingly beautiful artwork for novels that would’ve been a once-a-year project before, but now in less than a week.

Another artist has mostly specialized in custom paintings on objects other than canvas. She would do several mockups for potential high-paying customers before they’d ever agree to the commission. She can now upload her own artwork and combine it with her ideas and have 5 potential options in under an hour.

That used to be at least half her time on each project.

My point is, instead of dreading what is almost inevitable — that AI will be the new tools of artwork — why not be among the first to master the new tools and use it to be far more productive?

After all, if not you, it will be someone else.

Takeaway

I used to think of AI as being a replacement. Now I think of it as a tool. I grew up as one of the lucky few earliest kids with a computer the day I was born in the early 1980s. Because of that, I can type 100 wpm if I need to.

Ask me to write this essay on paper, and my hand would’ve cramped up three paragraphs in. I can only wonder what that same concept means for a child born with more advanced AI programs in a few years.

What will they create that we haven’t even imagined could exist today?

As for me, I can now take the boring part of writing out of my bland Googleable articles (Not here in Substack, don’t worry. Well, except this one.) I can take inputs, heavily edit them, and add my own voice of humanity. I can then wander over to MidJourney, and put in a weird pun related to the title, spend 10 or 20 minutes iterating, and come out with hopefully something that will make you laugh.

A laugh I wouldn’t have been able to create, nor afford, nor wait for even 1 year ago.

I hope that someday soon I’ll be able to do the same with novels, or books, or snarky cartoons where I can make fun of our evil overlords and not lose a moment of sleep over it. But I want to do those in my own voice, using my own wacky sense of creativity, to make worlds and images that have only ever existed between me and the universe of reality that somehow exists in my brain.

I simply want to create.

For me, being able to put my thoughts onto screen in an even more visual way seems nothing short of amazing. And hopefully, hopefully, you’ll think so too. Even if I used advanced computer software sitting on a lonely server somewhere out there to help me create it.

After all, despite my university’s best hopes, I’m a thinker, not a doer, but now I have a machine to be a doer for me.

And you can too, if you want it to.

A magical landscape

J.J. Pryor

For more weird stuff, come check out Pryor Thoughts.

Leave a Reply

Your email address will not be published. Required fields are marked *