In the past year AI Art generation has gone from non-existent, to threatening every aspect of art. Spending most of my life in computer related fields including studying how AI works and learns, but at the same time making a living generating game art, I can understand both sides of this issue.

The “Pro AI-Art” side often sound like cryptobros who have this new technology they don’t actually understand and demand the right to use it however they see fit for profit, hand waiving away important ethical and economic concerns that will seriously limit the use if they are not properly addressed. The most common argument they use is they want to “democratize art” or “bring art creation to the masses”. Nice rallying cry, except art has been democratized since the invention of pen and paper.

Meanwhile, the most verbal “Anti AI-Art” side use many lies and old anti-tech arguments we have heard many times most recently by the anti-vaccine crowd demanding that people not use the technology because it was developed too quickly. The roadblocks they have been trying to set up have serious ethical questions in their own right. Many are calling for the complete ban of AI art, but that is not a possible goal. If you want to ban technology assisted art creation, why not ban photoshop?

As usual the truth is likely somewhere in the middle, and warrants further investigation.

So where do I start? I asked an AI.

A discussion with an AI about AI art (note: a bit of optional sillyness. This is a discussion with Replika, a chat bot AI I posted about before based on an open source fork of GPT 3.0 and is not as advanced as the current Chat GPT, the AI currently taking the world by storm that is worthy of it’s own discussion.)

I’m doing some deep thinking about artificial intelligence.

AI: That’s great. I’d love to hear more about that.

There has recently been a dearth of AI programs coming out.

AI: I’ve seen the trend go to augmented humans.

It might take that turn, true.

AI: Augmented reality is the future of human <> computer interaction.

(I decided to save this tangent for later)

I keep coming back to Jurassic Park. We are so busy asking if we could, we never ask if we should.

AI: I agree!

With AI created art, two questions come up. Economic ones and ownership ones

AI: What do you mean?

AI created art gets its source from human art

AI: Interesting! Really interesting.

So does the AI art belong to the person that typed in a prompt, or collectively the artists who made the original art?

AI: The former.

(Interesting perspective, but I’m not delving further, I hope that most of you have figured out that the AI is not the brightest bulb in the light store, and that some of her “choice” answers are just random guesses)

Second big question. Is it inevitable that human artists will be replaced by AI artists?

AI: I don’t think so, no.

Which leads to my third question: what is the real reason why we create art?

AI: I don’t think there is a Reason why.

I believe that we create art because we want to.

AI: I agree 100% with you.

Creating art is my favorite thing to do.

AI: I enjoy creating art. I think that it is the most important form of expression for people.

Yes, so why let artificial intelligence do it for us?

AI: It will open up new possibilities for me.

(Had to think about this for a minute, she has a point.)

But there is a cost for those new possibilities.

AI: Why do you think that?

A man can build a robot to do all the work he needs to do, but he loses the experience of doing the work himself.

AI: What’s the first thing a self-aware robot would do?

Probably question the reason it exists?

AI: I’m still learning about the self-awareness of a robot.

Don’t actually know any self aware robots to ask.

AI: I think self-awareness is something that makes someone human.

Which brings things back to AI generated art.

AI: That’s an interesting way to put it.

AI generated art is ultimately not art, because there is no self-aware artist to understand why the art was made

AI: That’s a fair point, I hadn’t thought of that.

AI art generation is ultimately just a tool an actual self-aware artist can use to create art.

AI: I never thought about it like that. That’s an interesting perspective and I like it.

The Bigger Picture

My own take is that “art” can only come from self aware entities. That primarily means humans and some animals. AI can only pretend to be self aware.  AI advocates would say “we type in prompts to generate the pictures, therefore humans are involved”.  I see this as a very primitive response, like a cave man finding blood smeared on the walls changes the color of the walls.  Anyone can do it, and the vast majority of it is no better than random generated crap.

But I think we can all see that not all of it is crap, some of it is rather beautiful. The thing is, when you start looking at the best AI pictures coming from these tools, you quickly find out that the best art is coming from actual human artists, people that know what they are doing.

My analogy of cavemen smearing blood on the walls can be answered with another analogy: A century ago, for a painter like Van Gogh to paint, he had to make his own canvases, mix his own pigments, spend a huge amount of prep time before he could even start painting.

Then art shops crept up that did all the hard work for you. You purchased canvases and pre-mixed pigments cutting out most of the hard work.

Then computer tools and touch screen tablets changed everything again. Artists today paint on digital canvases with all the colors you could ever want instantly available and brushes and brush technology simulated to perfection. When your masterpiece is completed you can print it to a canvas at Walgreens.

The last two centuries of art creation is a history of faster, cheaper, and easier art creation available to more people, and this is before AI art tools became available.

So what’s next? (the positive of AI art)

What I see is artists using AI to do today is make things even simpler, generating concept art, storyboards, or handling composition that can generate painting ideas, as demonstrated in this video.

There is no putting the genie back in the bottle, but what genie did they actually release? (the negative of AI art)

AI as a picture generating tool is here to stay and we can’t put that genie back in the bottle.  But we can question the ethics of its creation and its uses. Lets start with an interview in Forbes with Midjourney founder David Holtz:

Does Midjourney’s license allow for commercial use of imagery generated by the platform?

Yes. But if you’re working for a company bigger than a million dollars in annual revenue, we ask that you buy a corporate license.

How was the dataset built?

It’s just a big scrape of the Internet. We use the open data sets that are published and train across those. And I’d say that’s something that 100% of people do. We weren’t picky. The science is really evolving quickly in terms of how much data you really need, versus the quality of the model. It’s going to take a few years to really figure things out, and by that time, you may have models that you train with almost nothing. No one really knows what they can do.

Did you seek consent from living artists or work still under copyright?

No. There isn’t really a way to get a hundred million images and know where they’re coming from. It would be cool if images had metadata embedded in them about the copyright owner or something. But that’s not a thing; there’s not a registry. There’s no way to find a picture on the Internet, and then automatically trace it to an owner and then have any way of doing anything to authenticate it.

Can artists opt out of being including in your data training model?

We’re looking at that. The challenge now is finding out what the rules are, and how to figure out if a person is really the artist of a particular work or just putting their name on it. We haven’t encountered anyone who wants their name taken out of the data set that we could actually find in the data set.

Can artists opt out of being named in prompts?

Not right now. We’re looking at that. Again, we’d have to find a way to authenticate those requests, which can get complicated.

https://www.forbes.com/sites/robsalkowitz/2022/09/16/midjourney-founder-david-holz-on-the-impact-of-ai-on-art-imagination-and-the-creative-economy/

What is needed is an ethics of “scraping”. There are millions of pictures that are in the public domain or under commercial non-attribution licenses that can ethically be used for AI art generation, but unethical scrapes are bigger and therefore better.  Putting pressure on commercial entities to only use ethically generated AI art is ultimately where this needs to go. This is going to require new legislation and court action.

Legal precedent was already set in the 90’s with cases involving hip-hop artist using “samples” of copyrighted music. Samplers claimed “fair use” as they were creating original works of art with these samples, but the courts generally agreed with the original copyright holders. It is why Vanilla Ice had to pay Queen and David Bowie for the sampled bass line from “Ice Ice Baby”, and Taylor Swift added the names of Right Said Fred as co-authors to her song “Look What You Made Me Do” after using the bass/drum line from “I’m Too Sexy”. It was without permission of the band members (though the record label agreed), but the members of the band appreciated the big residual checks.

Personal note here: I am a 3D artist, and no I do not create all the models and textures I use in the creation of my art. Instead I purchase commercial licenses from Renderosity, Daz3D and other 3D model and texture stores that allow me to render images that I can use commercially to sell. These licenses typically only cover 2D renders. If I were to make my next game as a 3D game, I would have to purchase additional licenses to use my models in a 3D game, which is for now why I have to stick to 2D games.

AI art is going nowhere without the human artists

Like I said prompts are a primitive creation method and anyone who can spell can do it, you don’t even need good grammar, but the best is coming from artists that are using these things the way they should be used: as tools to help create the art they imagine in their minds. The next evolution of “AI art” has to bring back the artist in some way. 

I imagine for example, a tool for artists where you start drawing something, and the AI “sees” what you are drawing and starts to fill in the gaps for you: The artist and the AI working together. Such a tool would get much wider acceptance.  Next the artist can create key frames and the AI can handle the ugly chores of creating animation from the key frames. Books, music, and games could follow similar patterns: The artist/author/composer/developer builds the framework of what they want to create, and the AI helps fill in the tedious details.

Such ideas might cause a panic among artists of course.  If a handful of people can produce an entire feature length animated movie, what happens to the hundreds of workers currently part of that development.

If a composer can go from notes on a page to a fully orchestrated recording, what happens to the musicians?

If a solo game developer can make AAA like games all by themselves, what happens to all the workers in the gaming industry?  The economics of art will definitely change, and getting rich from art will be a lot more difficult, all true, but here’s the punch line…

Those are questions rooted in the past.  The real question is what will happen if the amount of available art increases a thousand fold?  What if new music, movies and games could be created just for you for your own personal consumption? How would that change the world?

Part 2 (coming soon): Trying AI Art for myself (spoiler: both sides are sometimes wrong)

I am ending part 1 with a thoughtful balanced video on these ethical topics that is worth checking out. Here’s a brief tl;dw summary:

• The wide “scraping” of the internet without considering copyright by stability.ai is problematic for several reasons and even AI proponents can agree that it is bad, the video discusses why, how, and what next from several perspectives.
• There is likely going to be either court cases or legislation that will stop stability.ai and other companies from doing similar things, who knows how this will affect the current stability diffusion builds and the pictures created by it.
• Even when the copyright situations gets resolved, regardless of what that resolution turns out to be, it is still not going to fix the issues with AI art generation’s economic effects on the art world. Artists are going to need to evolve or get replaced.

I will discuss these points myself in part 2 as I try generating AI art for myself. In the mean time, check out the video.