What’s The Deal With A.I. Generated Art?

This article is about the recent trend of visual art that has been created or assisted by A.I. (artificial intelligence).  If you’re not familiar; most of these apps are “text to image”, meaning that the user types a text prompt (such as ‘a black stallion galloping through a meadow’), and the app uses advanced algorithms and graphics processing to create a complex image based on it’s “knowledge” of what common objects look like.

Full disclosure;  As recently as a few months ago, I was not a fan of A.I. generated art.  Not only did I consider it “cheating”, but I sincerely wasn’t impressed, at first.  My first exposure was when Wombo’s dream app first gained popularity online around 2021.  For the most part, the images looked very similar in style.  And although admittedly eye-catching, the wave of Wombo images got old very fast.  I started seeing other A.I. apps pop up, and while they seemed a bit more “serious”, I still didn’t agree with the general idea.  Something about it, and the thought of “dishonest” art created via algorithm did not sit right with me.  I won’t lie- part of it was the absurdity of seeing people take credit for visual art that was created by a computer, or the equally strange claim that writing the initial text prompt is now “the hard part”.  To be honest, I’m still coming to grips with these ideas, but surprisingly my attitude towards the matter has made a complete U-turn.  Yes, you read that correctly.  I now fully endorse A.I. art, even if the relationship is a bit contentious.  Let me explain how I came to change my mind on this topic.

I guess you can say that I’m a man of scruples- if I’m going to talk shit about something, I absolutely have to have enough first-hand experience with that thing to have an educated opinion.  I mean, there’s nothing worse than someone who ignorantly bashes an artistic tool, trend, or technique that they haven’t studied themself.  Since I am a graphic artist, and I don’t want to be “that guy”, I decided that I would immerse myself into the currently available A.I. art tools, and really see what they’re all about.  I picked out the top three A.I. text-to-image tools, and thought I would dedicate some time and money to trying them out extensively.  I had to see them up-close for myself.  Based on what other artists seemed to be using and the results they were getting, I settled on Midjourney, DALL-E2, and Stable Diffusion.  At this point, I’ve spent a considerable amount of time with all three.

Before I get too lost in the weeds with details, intricacies, and deep philosophical rants regarding these art tools, I should probably explain why I changed my tune regarding art created by artificial intelligence.  Perhaps the biggest reason is; because it’s here and it’s not going away.  A.I. art is like a genie that’s been let out of its bottle, and now you can never put it back in.  It really is the truth, though; you can’t un-invent things.  If you ban guns, people will make them in their garage.  If you ban alcohol, people will distill it in their garage.  A.I. art is no different.  We live in a world where technology can’t be undone.  

With that said, going forward from here there are going to be two types of graphic artists in the world;  Artists who embrace A.I. as a tool, and the artists who don’t.  The first group (artists who use A.I. to some extent) will be much more suited to actually find work in the future.  They (the early adopters) will become the future leaders and trend-setters in the arts.  As A.I. text-to-image tools become better, and are more widely adopted, the most capable and in-demand artists will be those who have traditional skills AND are savvy with Artificial intelligence.  If you refuse to use these new tools to your advantage, it will probably be to your detriment. 

It’s important to point out here that the traditional “employer” of graphic artists, i.e. a person who commissions artwork (particularly for commercial use, such as a logo or album cover), does not care about anything other than the finished product you deliver to them.  For most graphic art that is created for commercial purposes, the person paying for the art (usually a business) does not care how many hours it took you to do it.  They don’t care how many years it took you to hone your craft, nor do they care about the “story” or “process” behind how the work was created.  They literally only care that you made the thing they asked you to make.  They want you to create their “vision”… they typically don’t care how you get there, as long as you get “their vision” correct at the end of the day.  

Here’s a little insight into what the process of creating commercial art is like.  You get commissioned from a business, let’s say in this case, it’s a band that wants an album cover graphic of a flying unicorn.  So you draw a few sketches of unicorns, and ask which one they like the best.  They go with #4, and you go ahead and start detailing that one out.  You send it back to the band, and they say they like it, but they ask if you can make the unicorn’s horn rainbow colors.  So you fix the horn, and turn it back in.  Then they tell you that they actually want to go with sketch #2 instead.  So you go back to sketch #2, and start detailing that one.  You turn it in.  They tell you that it’s good, but now they ask if you can draw a forest in the background.  You wish they’d have told you that beforehand.  The type of forest they are asking for won’t actually look very good, but they’re paying you to create their vision after all.  So, you painstakingly go back and add in the elements they want even though they don’t look good.

Now I have to ask… does that example (which is typical in the art business) sound the least bit artistic?  Is that making art, or is that something else entirely?  If you were the artist in that situation, wouldn’t you want a “magic button” that you could press which would give the client an image exactly as he was imagining?  Or, at the very least, would you not want a tool that could simply draw what the client is describing?  In my experience, creating graphic art for clients really isn’t artistic at all- you’re simply providing your skills (the ability to render a drawing) for payment.  And as already stated, the client only cares about the picture you hand them, and if it looks like what they intended.  They don’t care how you got there, or how much of an “artist” you are.  Trust me that 90% of people really don’t care about those things… They only care about the end product.  

Now don’t get me wrong… there are some people (10% of buyers?) that do care about the story, the provenance, the process, etc.  But these are art collectors, and not people buying are for commercial reasons.  These are people who legitimately collect art, know and research artists, and actually give a shit about why they’re buying what they’re buying.  For those people, it’s not just about the instant gratification of a cool picture.  But right or wrong, these types of buyers are not the norm.  Understandably, you can’t really expect commercial art buyers to think this way.  If you’re buying art to make something else look good, you are only going to care if it looks good.  

If you’re a fine artist with a reputation for doing a certain thing, then perhaps a sudden change to A.I. based art isn’t for you.  But if you’re a graphic artist, concept artist, or other type of artist who is mainly concerned with the “end product” then A.I. art will almost certainly help your workflow.  Especially if you’ve been tasked with creating visuals that must fit certain constraints, then A.I. might be a useful tool. 

Concept art in particular is going to be a niche that will be heavily affected by A.I. art.  I can see how concept artists might feel the pinch from this… Now, instead of being the person that takes a creative director’s “prompts” and draws them into life, the director can just use A.I. to generate images (hundreds if necessary) until he gets what he had in mind.  However, very specific types of art (like character breakdowns) can be extremely difficult to create using A.I., so the human artist is still safe in that respect.  I see A.I. generated art already popping up in a lot of places, such as the images that accompany blog posts, podcasts, and similar content.  Those types of media, because they are released often, usually need some sort of accompanying image, even if it’s relatively low-effort.  A.I. art is great in this situation, and effectively replaces the need for stock photos and artwork.

I’m not saying that this technology won’t be disruptive- it almost certainly will be to some extent.  The best analogy I can draw is to the creation of MIDI in the 1980s.  Back then, when digital synthesis first hit the scene, many musicians thought these products would end their careers.  The thinking was, “why would anyone hire me, a real musician, when I can be replaced by a programmed synthesizer!”… samplers caused an even bigger uproar.  In all fairness, I’m sure there were musicians that did lose gigs for those reasons.  When those technologies first hit, they were used a lot, and in some cases instead of “real musicians”.  But in music (like in all art), trends come and go, and “real” instruments never went away.  Today, we have room for everyone… There are still acoustic horn, piano, and cello players, just as there are synthesizers that emulate them.  

What then, is my final take on A.I. created art?  Are these tools legitimate?  Are they “assisting” the artist, or doing the work for them?  Like I already mentioned, they’re here to stay whether you like it or not.  There are certainly artists who are using them legitimately, and there are also total posers who are using them to pretend that they created something they did not.  I think in the future, things will shake out a bit more and the general public will be a bit more savvy at spotting legitimate art from the noise.  I also think that the “real” artists will find ways to incorporate A.I. as a legitimate tool, and take the results farther than anyone can imagine.