The past few weeks have brought discussion about the influence of Artificial Intelligence in various artistic environments. I’ve posted about AI text generators before, and I spoke to my musician brother about AI in music—which may not be quite as robust quite yet, but which could include things like automatic drummers and Other Synthesizing Things. The latest wave is, of course, about visual arts.
Specifically tools like Midjourney (which is a discord-resident tool) and Dall-E. There are several others, too. Just use google, and your mind can boggle. Stable-Diffusion is probably number three on the list as far as I can tell. It has a coarse online version. (which I used for the somewhat uninspiring image at the top of this post)
One recent news piece focused on a “prompter” who used the tools to win an art contest.
I’ve also spent more than a couple hours watching video of artists react to the tools, the most interesting and in-depth of them being something renowned artist Dave McKean participated in.
The underlying vibe therein is the concern of whether this is the end of art, or just the end of artists.
This is an interesting question. In this case it’s quite relevant.
I say that because the work coming from these tools is truly interesting. I mean, just fiddle around for a bit and you can find a LOT of amazing, eye-catching stuff.
Theoretically anybody with a little time can sit down and start tossing around prompts that get them some great art. Full disclosure: beyond that little Stable-Diffusion demo, I have not yet fiddled with it myself, but I’ve seen other indie writers doing so. The result has been quite buzzworthy.
Add in that it appears that art coming from AI generated tools cannot be copyrighted—meaning theoretically I or any other publisher, independent or traditional—could use anything we created with the tools, and it would seem that the price is really quite right, too. It’s important to note, though, that this has not been tested in court, so really, my hackles go up when it comes to the idea of being first in line there.
I have a few better things to do than spend a few weeks in court.
Of all these newsie things I’ve been digesting on the topic, the most interesting to me are ones in which artists are thinking through the situation. The interview with Dave McKean I linked above, for example, is a level-headed look at the future made by people making their living around the world of art. It explores the concept of quality and effort and the capitalistic aspects of commercial art—i.e, how much art might cost going forward, and who actually does it.
Their answer: things are going to change, and they are going to change fast.
But exactly how they are going to change is only partially discernible.
Minimum Viable Product:
Let me wander off the path and talk about something called a Minimum Viable Product.
The Internet/software world arguable created the whole concept, or if they didn’t create it, they took it to its fullest example. The wild, wild badlands of the early and mid-90s Internet landscape was all about MVPs. That was a decade or so in which it did not matter how buggy your product was as long as it mostly did its job and you got it to market first. Getting to market first with something that provided something that kind of worked created multi-millionaires and billionaires, getting to market second created very unhappy programmers.
The independent publishing world went through that phase in the first decade of its existence, or is arguably still going through it to some degree. Hell, it’s arguable that publishing has been a great example of MVP since its earliest days as a mass product. Pulp fiction, anyone?
The whole concept of Minimum Viable Product fits the publishing model because (1) quality is in the eye of the reader, and (2) readers are who they are. Perhaps the old pulp fiends were not exactly James Joyce or the Brontë sisters, but a lot of people liked reading them. Or if that gets your knickers in a knot then, sure, Harlequin romances are not the most daring of literatures, but try telling their readers that the stuff they love isn’t “good.”
In today’s world, one of the big 20Booksto50K guys, Michael Anderle, has been a public proponent of the idea, and whether you like the idea or not he’s ridden it all the way to the bank.
Naturally, he comes from a marketing background. (grin)
He and a few others make big splashes by throwing quickly written material into the world and marketing it hard to people who like it. He’s doubled down on that in the same way James Patterson has, creating shops of other writers doing the same thing under his name. Ethics of certain practices aside, the point stands.
I use Anderle (who I have not met) as my example, only because he’s been open and unabashed about leveraging the concept of Minimum Viable Product while talking about his work. But it’s worth noting that there are a LOT of others using similar philosophies, though they employ the phrase “write to market” more than Minimum Viable Product.
Are these books good? Well? Fact is that they have lots of readers. People read these books because they love them. The stories captivate.
Therefore, they are Good.
Or at least Good Enough.
Why Does this Matter?
We were talking about Artificial Intelligence and whether it might put artists as we know them out of business, right?
At question is exactly how good is “Good Enough” when it comes to art and illustration.
As a rule, these artists I’ve been listening to can look at AI generated product and see that it’s at least partially mechanical—essentially derivative. Even if the images are eye-catching and stunning, they say the can see a difference. Kind of. This makes sense to at least some degree because AI generated art is created by a machine learning algorithm that has been “taught” by assimilating billions and billions of pictures and compiling its own rules. There will not be any real creativity happening inside the machine (as we know it today, anyway). The machine, if well prompted, is technically more proficient than a human, perhaps. But you’ll not see it purposefully breaking any rules to create something new.
[aside: then again, there’s Bob Ross and his “happy little accidents,” right?]
“Good Enough,” though is an interesting concept.
Here’s a side example from my last weekend.
I am in the process of working on a fun new project with my brother, who is a musician. As will be revealed fully soon, we’re working in multi-media, something I’ve enjoyed so far because we each “own” certain pieces of the project. For example, I manage all the text, and Jeff responds to it. Jeff owns the music—as well as the final sound recording cut on text we record.
At one point, I created a chunk of audio by reading text on my computer. I then went into my sound editing stuff and made it “better.” When I handed it to him, I thought it was pretty darned good. To my ears it was as good as most things I listen to, and to be sure I am not a neophyte listener. I consume all sorts of podcasts and audio stories. This would have played in that environment. It was “Good Enough.”
Jeff, though, who is much more attuned to sound and sound engineering than I am, immediately picked out things to improve. He tweaked this and adjusted that, and next thing you know it really did pop.
What Jeff handed back was considerably more robust. Quite professional.
So, yeah. Just what is “Good Enough” and how will it apply to us when it comes to art? Particularly art in the world of publishing, where the purpose of a piece is “simply” to sell the book?
The real answer is I don’t know, and neither do you.
Everything is just too new to tell anything. Still, I have a few more thoughts.
Thought 1: MVP = Gut Punch?
I admit that as an “Artist” (har!), the idea of applying the “Minimum Viable Product” concept to creative fields bothers me.
Of course, the flipside is that even the most gifted writers of all time—or any other artists for that matter—can be notoriously bad judges of the quality of their work. So, at some point we really do need to fall back on this idea of MVP and say “this is at least As Good As I Can Make it,” release it, and move on to the next story.
Maybe that’s not exactly the same thing, but it’s in the ballpark.
At some level, it’s not only helpful to accept the concept, but is just good for both business and mental health.
Thought 2: Has Video Killed the Radio Star?
When I was still a teenager, a band called the Buggles put out a song (that became the infamous first song ever played on MTV) titled “Video Killed the Radio Star.”
Its lyrics are quite intriguing today, and the conversation at the time was all about how MTV changed things. And it was true that there were changes.
But, no, it turned out that video did not kill the radio star.
Not exactly, anyway. In fact, relatively soon as evolution has things, MTV became more than a bit of a joke for playing mostly NOT music. But the radio star has been mostly killed today, not really because of MTV, but because of the Internet and streaming.
Let’s talk for a moment about that term “Radio Star,” shall we?
Ultimately, that term sounds like it’s about the artist (the Star). But I say here that the term was really about the medium (Radio).
There are, after all, still some massively huge “stars” in the music world. So the “Star” has not died. Instead, the “Star” has been agile, and has moved on. Similar to the Hollywood shift from silent film to talkies, creators either shifted with the media and distribution channels or, yes, eventually faded, or at least diminished in their ability to make big impacts on the larger world. Music is still on the radio, of course, but the listener base in that distribution channel is considerably smaller.
Instead folks are getting their tunes through Internet streamers. My wife and I, for example, are all-day, every day listeners of Radio Paradise, a really great independently programmed channel that plays widely eclectic mixes of music by a wide array of artists.
Aside: Here’s an interesting little thing that gives a very high-level breakdown of how musicians get paid for streaming media.
So, neither Video nor streaming killed the Radio Star. Both of those changed the business model that the industry ran on, and therefore the processes of becoming a Star. But they did not kill the idea. Assuming the Artist Formerly Known as “The Radio Star” was nimble enough, or at least able to look forward and be willing to play in the new environment, they are not dead.
Thought 3: Just What is an Artist?
Still the music is being created differently than it once was.
Everything has changed, really, specifically including the tools artists use—the mechanical aspect of how sound itself is literally created and captured.
Auto-tune completely changed what it means to need to be able to “sing,” for example. Artists that at one time we might have called “sound engineers” are making songs that sometimes use instruments, and other times come completely from within a machine’s components. Sampling methods are now mainstream.
There are still guitar players and drummers and whatever, but a whole world of technological tools that are not traditional musical instruments are being employed today that did not exist even ten or twenty years ago. And the wielders of those tools are … wait for it … artists themselves.
But … um … yeah.
There is, for example, a human skill that I don’t have and that my brother does. That skill allows him to create a better sounding voice by using a doodad that does not itself make sound. There is also a level of experience that he’s built over a life spent making sound that I don’t have—and that experience and focus allows him to realize the difference between “Good Enough” and “Exceptional” when it comes to sound. That means he can twist and turn the knobs on that little machine is ways I cannot, and fairly rapidly come to a great sound.
When he twiddles those knobs, he’s creating art.
I suspect the same will be true here to some degree.
There is, for example, (and as I understand it), a human skill to creating prompts that then create images. That human skill is already being marketed, just as a street artist is peddling his or her art on Fremont street when I walk down it. It’s an interesting concept.
The work itself is, of course, totally different. Rather than brushes or mouses, the “prompter” is literally creating images with words. The person sitting at the computer, twiddling prompts is not going home smelling like oil paints, and they aren’t spending hours in the studio perfecting the use of different watercolor brushes—but, then again, how many book cover artists today are creating actual physical art rather than working completely with digital creation tools? How big of a leap is it from making art with a mouse, versus making it with a keyboard? “Prompters” are certainly creating something, after all.
And it’s the eye of the prompter as to whether something is “Good Enough” for their purposes. It’s the “prompter’s” artistic taste that decides when something is “done.” In the most complex way, the AI is simply a new instrument. A new brush. A new synthesizer. It’s still the “prompter” (or the art director, or the sound engineer, or the marketing head, or …) who decides whether the result is marketable.
So, will AI generated images “replace” images created by human artists?
I think the answer will be “Kind of.”
People will always value art made by other people. I’d guess fine art in particular will not change terribly much. I mean, spend a half hour in the Monet room in the Chicago Museum of Art and tell me you’d feel the same way staring at Midjourney work and I’ll call you either a liar or (at least) tone deaf to what art as a human endeavor means.
But the question here is really about commercial art. Book covers and album covers and graphic design and other such stuff. Art that’s purpose is to sell another product.
I think a lot of that will change. And I won’t be surprised to see that change soon, too, because so much of it already meets the whole Minimum Viable Product standard. So much of it already catches the eye. A lot of it that I’ve seen could already quite easily be pushed into a book cover and garner the attention the publisher is trying to garner. Assuming the copyright issues get actually hammered out—which they will—I can see traditional publishers (and indies, for that matter) hiring the best and brightest “prompters” to create images for them because those prompters will almost certainly be cheaper than the people we know today as artists.
I can’t image that won’t happen.
At first (as in Real Soon Now) I can see these cheap and readily available generators changing the independent business model by simply raising the bar on what it means to be Good Enough to provide cover art.
In the early days of independent publishing, for example, covers could look pretty ugly and still sell well. Now the bar is raised because most serious independent publishers have understood that value of “Good” covers, and either built a skill themselves or hired artists to make themselves more competitive. Crappy covers no longer cut it.
These Midjourney and Dall-E images are really quite good. I’d guess they will very rapidly drive even more indie writers away from doing their own unless they have actual skills. And, in fact, I’d expect there will be an entire cottage industry that will spring up built around helping the masses of independent writers learn how to use the tools to make interesting art. In that longer run (maybe a year?), I can see the ability to use Midjourney or Dall-E or Stable Diffusion as being an important part of the independent publisher’s toolkit.
But, for the near term I figure momentum exists that means there will be human artists in the loop much more often than not.
I suspect traditional publishers will wait to get into using AI until the copyright thing is dealt with—and once that happens they will get in big.
And similarly to the example of music packagers—independent musicians and small labels will adopt the use of AI art early (because it’s cheaper), and bigger labels will jump in later because they are cautious and because it takes a while to turn an aircraft carrier.
That’s my view, anyway.
I don’t have a real timeline in mind. I suspect it’s earlier than later, but I’ll keep the definition of earlier close to the vest for now.
The main point I walk away with from this past couple weeks, though, is that, as an independent publisher of my own work, I’m keeping my eye in this closely because the times are certainly changing.
I’m guessing I’m going to be needing to learn how to use a new tool pretty soon.