Prompt. View. Repeat.
Discover more from Iterations
Prompt. View. Repeat.
There’s no ghost in the machine. Yet.
I feel like I should apologize in advance in case this comes off as a bit of a rant, but it’s something that I’ve been thinking about for a while now and I just need to get it out of my head. I’m talking about AI.
You don’t have to look far to know that AI is everywhere and it doesn’t seem to be going away anytime soon. It’s in the news and on social media. It’s what your colleagues are talking about when you’re at work and what your friends are talking about when you’re not. And no matter what you do for a living or for recreation, it seems like AI is going to either disrupt it completely or at least affect it somehow.
I’ve been following the development of some of the more popular AI platforms from the wings for a while now. I’ve tried ChatGPT, and I recently did an experiment with Midjourney and Stable Diffusion to see how they would handle prototyping ideas for a personal project I’m working on. Even though the results weren’t exactly what I wanted, they got me thinking about what I did want in a different way and ultimately gave me some new things to think about and iterate on in my sketchbook. While I can absolutely see the potential power of AI, I can also see some of the potential cost—and that’s what concerns me.
And when I say cost, I’m not talking about the economic or societal cost that someone like Geoffrey Hinton—who recently resigned from Google’s AI team—is talking about when he said that we’re at a “pivotal moment” with AI. I’m not nearly intelligent enough to even comprehend that level of informed concern. And I’m not even talking about the potential moral or ethical concerns I have about being able to generate images of whoever you want doing whatever you can imagine. In this Iteration I’m talking about the creative cost and the existential cost to creators who are moving to AI for their art-making.
Take photography, for example. AI is dominating the conversations on Twitter, Substack, and Instagram, and to a certain extent, I get the excitement. The results that people like Nick St. Pierre and Julie Weiland are sharing on Twitter and some of the stuff I’ve seen on YouTube lately are really incredible, but it’s not photography. Not only that, I think it’s missing the point of photography—or any art-making—and that is choice. Creativity is all about making choices and then reacting to those choices. And before I get too much deeper, I need to remind you that this is just for me and the way I see it at the moment. You may disagree and see it completely differently, but for me, photography is about the challenges and the joy that comes with looking at the world through a viewfinder—or even the back of my phone—and choosing a composition that works for what I’m trying to communicate. That means making real decisions about real subjects and real light in the real world, not merely prompting machine-driven interpretations of those things. It’s also about accidents, happy and otherwise. Maybe I snap the shutter 1/250th of a second too late and didn’t get the exact expression that I saw or wanted, but the photograph that I captured ends up better for it. I captured an interim expression—an expression between the expressions—and that’s what made the photograph magical. At the moment, you can’t describe that to an AI, and even if at some point you can, you still miss out on the experience. AI removes the “humanness” from the creative process, which is one of the most personal and existentially satisfying processes we have as human beings. And for what? Efficiency? A lack of skill or resources?
Adrianne and I were talking about some of this and she showed me a cartoon with a caption that read, “AI is supposed to cover the boring stuff to free us up to make art, not to make the art itself,” which I absolutely agree with. She also made an interesting observation around the potential motivation for using AI to make images. She said, “The physical experience of making apparently isn’t valued or at least it’s not the motivation or driver for a lot of the folks using AI to generate imagery. They’re effectively just writing a program to be run.” I thought it was interesting that she used the word “generate” rather than “make” because that’s really what it feels like at the moment. When you enter your prompt into Midjourney, it generates four variations for you to either keep as-is or use as the basis for it to generate additional variations. But in each case, the machine is making the decisions and removing the why from the maker.
I’ve spoken to art professionals from conservators to curators and the one thing that nearly all of them have said is a prerequisite for Art is transformation—and that could be a transformation of the artist, of the audience, or of the materials. AI effectively removes transformation. You weren’t inherently transformed by writing a more descriptive prompt. You didn’t get any better at the craft of making images, and you didn’t learn anything deeper about yourself. All of the discovery about the world around you, about another human being, about yourself, about whatever it is that happens going through the process of making art is lost in AI. At the moment, it’s merely a transaction—prompt, view, repeat.
I was talking about all of this with Sean the other day. I brought up the fact that all of these applications are scraping millions of images online without permission in order to create the datasets they use as their source material, which is basically theft. In fact, there have been a number of class action lawsuits brought against AI platforms like Midjourney and Stable Diffusion, claiming copyright infringement. “Can I ask you a tricky question then?” he said. “What’s the difference between what AI does and what you do with your mixed media work? Both include the work of others without giving specific source credit in the presentation of the work.” It’s an interesting question, and frankly, I’m still sitting with it.
I think it comes down to the fact that I’m making thoughtful, intentional, and human choices about how and where I use the materials so that they help to communicate a particular narrative—and that rarely means using any of the ephemera I find as-is. Instead the images are torn, scraped, or buried under other layers of other ephemera and are transformed from what they once were into what I need them to be. So I think maybe it comes down to intent, which is lacking in AI.
AI is still in its infancy and whether or not you currently use it, I think it will affect everyone eventually, either directly or indirectly. Since I’m always going to prefer to make my own creative choices when it comes to my art, it’s not surprising that I’m not an early adopter. That said, I am a curious onlooker and I’m eager to see where it goes and how it affects art-making and creativity in general. It will be interesting to see how some of the current court cases shake out, since they are challenging the interpretation of “fair use” and what constitutes “transformative” or “derivative” work. Depending on which way these cases go, there could be dramatic consequences on how images can be used and shared moving forward.
QUESTIONS
What do you think about AI? Are you an early adopter? I’d love to hear what you think and how you’re currently using AI in your creative process. Leave a comment or email me at talkback@jefferysaddoris.com.
LINKS
Microsoft Designer - Stunning designs in a flash
AI Art Generator – Adobe Firefly
DALL·E 2
Midjourney
Stable Diffusion Online
ChatGPT
Rick Beato on AI
Generative AI Has an Intellectual Property Problem
Prompt. View. Repeat.
The fundamental problem is that we only get better by *doing*, but our laziness wants to get better by *having*. As I get older, I realize the importance in savoring that process. The only way to "having done" is through "doing". So I try to enjoy the slow progression of working on myself and my skills, even when it's hard and my results are pitiful compared to what can be found online.
But the bigger world shaped by social media, feels like it's pushing in a different way. So be it. Let them have it.
I don't mean to denigrate those who are actively exploring AI art, but the cheap commodification of AI products is pushing me towards decidedly reactionary stance (back to pen and paper!).
Your mention of “transformation” as a pre-requisite for Art provides (at least me) a really helpful framework for talking about AI and its impact on creative endeavors, much more so than the other debates around whether AI can “produce art”. Of course it can, insomuch as it can produce compelling visual canvases that people will respond to, both emotionally and commercially.
But whether the creator was in some way transformed via the making of the art, and whether that transformation lends those pieces more weight, is really interesting. In a way it intersects with your ongoing discussions around process, and how for many the process is more important than the final piece. In AI the final piece is really all there is, since the “process” from a creative standpoint is akin to microwaving a meal. It can taste good and provide sustenance to the consumer, but you as the “chef” put nothing of yourself into it.
My initial fascination with MidJourney was that I could throw a string of words together that conveyed some mental image I had and watch as the AI converted that into often intriguing results, as if it were somehow acting as a bridge between my subconscious and the world, making manifest thoughts and emotions that I otherwise could not express. And there lies the rub in terms of transformation - even if the resulting image was an accurate representation of some inner part of me, did I grow in any way or learn anything by its rendering? Nope. I just had the Matrix suck it out of me.
Now, if AI could produce output that caused me to reflect in ways I hadn’t before, and then channel that reflection into another creative expression, or a new way of interpreting the world, then would that count as transformative? In other words, if AI prompted me, would that be a collaboration worth pursuing?