Brendan Dawes on AI and Creativity

A conversation with this acclaimed digital artist about how he incorporates AI into his work, and about the future of creativity in art, film, and photography.

Brendan Dawes, “Black Mamba’s Revenge”, 2020

Brendan Dawes is a UK based artist using generative processes involving data, machine learning and algorithms, to create interactive installations, electronic objects, online experiences, data visualizations, motion graphics and imagery for screen and print. I first met Brendan when we were on a panel together at SxSW in 2008, and it’s been fun to watch his progression since then and collaborate with him on several projects. In 2020 he began using AI as a creative tool for various aspects of his art. I’ve been thinking a lot about how AI will impact creatives in the future, and I was interested to hear Brendan’s take. I’ve included a few pieces of his work within the conversation, but definitely check out his Instagram feed for more.

Gary:
So Brendan, what are your impressions of all these new AI tools, and how have you been integrating them into your work?

Brendan:
Well, I've always worked in a way that is very much about the idea of combinations. I think that's because of my background when I used to make like, breakbeat albums and remixes, when I was in Manchester training as a sound engineer and playing about with music and Akai samplers and all that. My heroes were Coldcut and Steinski, people like that who were chopping songs together from things that already existed. So I see the correlation with AI being kind of similar, at least in the way I use it, in that my process is not just one thing. The way I'm using AI is not just typing a prompt and then outputting a thing and going, "I just made that." Some people are, but I can't do that. Because... I don't know whether it's that I don't think it's complicated enough, or it's just too easy, or I don't feel I've done anything, you know? So for me AI has to combine with other things that I've developed or done. My hand has to be present in it, it has to be visible in the work. Which is why it's just me here working <laughs>, there's no one else.

Gary:
What are you using, Midjourney? Is that what you've been describing?

Brendan:
No, I'm using Stable Diffusion. So I've used a few AI things over the years, like my very first NFT in 2020, Black Mamba's Revenge, that did use a bit of AI, but I wasn't using AI to generate the image. In that piece I made this abstracted thing, this sculptural thing out of a fight scene from Kill Bill, Volume One. That came about because there was an AI technique that would strip out human figures in images. Not from films, just from still images. And I thought, well, if you fire a load of images from a film at it, it could strip out all the human figures. So AI didn't make the thing that I made, but I used it as a technique in the process.

AI motion capture data from “Black Mamba’s Revenge”, 2020

That was a few years ago. So now with Stable Diffusion, I find it more interesting. The other artists that I'm talking to that are using this stuff, the ones that are really into it, are the people who are installing this on their local machines so they can hack it and play around with it. Stable Diffusion is completely open source, where Midjourney and Dall-E, they're closed systems. So you don't know what's going on under the hood, you can't really play around with them at that level.

Midjourney right now is probably the most advanced from an image point of view, but I don't really care about that. I'm not really interested in a lot of the AI stuff you see that is recreating what's gone before, you know? Like make an image of a car or make an image of a fantasy scene or whatever it is. And that's all fine. But those things don't interest me because I'm not... I read a great quote the other day from Alan Moore about artists, that the power of art is to create an alternative reality. And that's what interests me with AI. I'm not interested in creating real life or things that look like real life. And I kind of like where some of these AI things are a bit squidgy and a bit off. So the way I'm using it now is that I've got work that I make in Houdini, I'm creating these forms that I've completely made by my own algorithmic procedural thing. But then with AI, I can take that and augment it with other thoughts and other avenues. Then if you do that over a series of images, you can create videos and things like that.

So, AI is like a collaborator. I think that's how I see it, it’s like a collaborator in the studio, that's got this massive brain. I mean, we can talk about intelligence and all kinds of other intelligences. You know, they've done incredible experiments using slime mold and to determine mapping patterns in the Tokyo Underground. It works as like a hive mind. So there's all these other things that have intelligence, and it's weird to me that we think human intelligence is absolutely the best intelligence there is, you know? <laughs> So, yeah, that's the kind of stuff that I'm interested in, really. I've been doing lots of research into it AI, watching lots of lectures, and everyone says it's the biggest shift in human evolution. And it's exponential. I think in two years we're gonna look back and go, shit! It's like when the iPhone came out and changed everything, how smart phones changed every aspect of our lives.

Behind the scenes video showing how AI motion capture was used, with collaborator Charlotte Edmonds:

Gary:
Yeah. And I think that most people maybe can't see that potential in it, since they're only seeing like the parlor tricks of this early AI, like ChatGPT and these text-to-image generators. But I'm interested in this idea of how people, I would say non-artists, experience creativity, and the idea that their perception of the value of creativity has shifted. Like, "Oh, I don't need to learn how to use a Photoshop or how to make photographs, because I can tell the AI what I want and it'll just do it for me." I mean, is it a good thing or a bad thing if kids don't necessarily need to learn how to draw or take photographs if they can just say a prompt?

Brendan:
Yeah. Well I think, if you're in the business of making bad hotel art, then find another line of work. Cause you're not gonna have a job!

Gary:
<laughing>

Brendan:
AI will just be churning that shit out forever! I mean, same thing with stock photos and stock images, you know? I think Getty banned photographers from uploading AI generated images. But the other side of it is that one day Getty might say, "We don't need photographers anymore to make images. We've got a room full of kids just banging away at AI text prompts." So I think these are gonna be interesting times. I think that is definitely gonna happen in some way. But it's like anything though, it's like, everyone can wield the pen, everyone can type words into a computer. But It doesn't mean that Cormac McCarthy is out of a job.

One day Getty might say, “We don't need photographers anymore to make images. We've got a room full of kids just banging away at AI text prompts.”

Gary:
Yeah, I know. I guess what I keep thinking about is that there is a benefit in the process of learning to become someone like Cormac McCarthy. You know, if I'm learning to write poetry—and this is more than just learning the craft or whatever— there is a value in me going through that process with my own feelings and my own thoughts. Over those years of learning how to write, there's a huge amount of self-knowledge that's also being gained in the process. And I think that gets leapfrogged by using AI, because if I can do a text prompt to write a poem about something, well then, you know, for most people maybe that's enough? They don't necessarily want or need to know how to actually write poetry. But there is a value in the self-reflection that it takes to write poems, or the...

Brendan:
The process.

Gary:
Yeah, the process is valuable, beyond just perfecting the craft, for your personal evolution. There's more to learning than just getting good at something. There is a personal benefit to you, and your intelligence. It's challenging yourself. And I'm not sure if, well, AI is just another tool, and now we've sort of leveled up and this machine is just letting me skip all these steps. Like when the Mac enabled everyone to be a typesetter, for instance. So is this just an exponentially bigger tool step?

Brendan:
Yeah, I think it is. I've seen discussions with people saying that we might have to have universal basic income because AI will make so many jobs redundant. You know, the real discussions about what does this all mean? When you see ChatGPT now, there's already another version which is even more advanced, right? And it is kind of mind blowing. Of course, I'm an optimist and have a kind of love for AI because it's showing me things that are beyond the limits of my own imagination. But you could say, "Well maybe you should get better at imagining things!" you know?

Gary:
<laughs> Yeah, that's kind of my point.

Brendan:
Yeah! But you know, the promise for me is being able to do things in ways that you could never have imagined. That that's where it gets interesting for me. It's almost like, if you take magic mushrooms, you are augmenting your brain to open yourself up to these other thoughts, right?

Gary:
You mean because our brains and our thought patterns are sort of already set, depending on our life experience, the way we’ve been educated, etc.?

Brendan:
Yes. So AI in some ways is a little like that, but that's absolutely frightening for some people. I think we're gonna see a monumental, exponential shift in things like writing and art. But I do imagine there'll be the opposite effect as well, where it's like, really super rich art collectors who only collect human, not synthetic art. And it's like a quaint thing, you know? <laughs> "But was AI involved in this?" "No."

Gary:
<laughs> "Is this art AI-free?"

Brendan:
Yeah. Or maybe, you know, the people who don't use AI meet on a Wednesday night in some weird basement or something. I dunno, I honestly don't know where it's gonna go. I mean, I make visuals, right? So we're gonna get a point where you can just type in ‘Brendan Dawes’ and AI will make my work. Now the point is: I haven't made it! You know, it might look like something I’ve made, or...

Brendan Dawes, experiments combining generative processing and AI, 2023

Gary:
It's something you might have made, or you could have made it or whatever. But it's not actually something that you made.

Brendan:
Right, you know, I don't want a copy. I don't want a Banksy-esque picture. I want a Banksy.

Gary:
Yeah, but… is that difference something most people give a shit about?

Brendan:
No. And that's the issue. But that's not really my issue, that's society's issue in the way we consume art, and most people's threshold of what is good is like on the floor, you know? <laughs> It is, you know? What people say is good, I'm like, really??? I think one caveat is that we are in an amazing time for the quality of the TV shows and series. Some of the shows are, like Severance, some of these shows are amazing! But I don't think people edit enough in their lives generally, or in what they consume. I think we need to edit way more, because we've got all this content coming at us. It's even gonna be more now.

And then you can make correlations with AI and when Warhol was screen printing. You could argue, “Well, he didn't paint those Marilyn's,” or whatever. He didn't painstakingly paint every one. He screen printed them and...

Gary:
He also copied those images from newspapers and advertisements.

Brendan:
Yeah. He copied those images of Elvis or Marilyn or, you know, the Campbell's soup tin or the Brillo box. I mean, with a lot of the stuff he was saying that anything can be art, you know, off the back of Duchamp and stuff. But yeah, I think if he was still around, he'd definitely be using AI for sure. You know? So that's why I don't get too hung up on it, because I think it's the next evolution of artistic tools and technology. There's always been that. I think it was in 1848, when you could first get oil paint in tubes? Which was like a revolution. You could go outside and paint. And probably back then people thought, "This is bullshit!" You know, "It's not real art!"

Gary:
Yeah, "If you're not mixing your own colors from pigments, you're a fake!" <laughs>

Brendan:
Well, you make a good point though there. So let's take Photoshop, as you mentioned. If you started using Photoshop when it was in version 3, there weren't people saying, "That's bullshit, we didn't have layers in version 2. We actually had to put the work in." It's like, but people, what's wrong with constantly advancing? It's almost like a Luddite response, oh, I really put the work in. Cause we didn't have layers in anything before Photoshop 3. Yeah, but wouldn't you have wanted them??? <laughs>

Gary:
Yeah, exactly.

Brendan:
I think what AI does is it ramps that up to an almost crazy level. You know, I just saw someone's quote that "prompters are the new coders". Now, there's a massive skill difference between coding and prompting, I think. But some people might argue there's not, you know?

Brendan Dawes, Collisions, 2023

Gary:
Well, I guess they're similar in the sense that you're giving instructions to a system, and getting results. And the people who are better at coming up with those instructions will get better results?

Brendan:
Yeah. Because I've seen people who've had to play around with some AI stuff, whether it's Midjourney or whatever, and they say, "Well, it doesn't work for me." They can't make it make anything good. So I dunno, it's certainly interesting times. Maybe in two years we'll have another conversation and I'll be on the street, and I've had to sell my house...

Gary:
<laughs> And that sums up this mixed public reaction to it. There's the sense of wonder at what AI can do, and then there's the fear of like, how's this gonna impact my work, my job?

Brendan:
Because, you know, in a year or so, you could you put all your film footage into an AI-driven system just type the kind of narrative you wanted and it would make your film.

Gary:
Yeah.

Brendan:
I think that's the next Adobe Premiere editing tool, right?

Gary:
I think that could definitely happen. I don't know if it's in a year, but rudimentary AI-generated video is already here, but that sort of capability is a few steps beyond still images. But it's coming and it's gonna be crazy.

Brendan:
Yeah. And then, if you want the editing to be in the style of Walter Murch or someone, you just type it in and it does it in his style, you know? At the end of the day, there are video editing styles that could be studied and replicated by AI.

Gary:
But it's the same thing that you brought up before. It's not actually Walter Murch. It's in the style of what he may have done, but it's not actually what he would do. You can mimic his style, but you can't recreate his individual intelligence.

Brendan:
No, no. And I think that's where the nuance is gonna be.

Gary:
But again, does anybody care about the nuance??? <laughs> Or is it like 1% of the people who really care? I especially think of kids growing up now, if you're a five-year-old right now, you don't care. You say you want a drawing of a bunny rabbit jumping through a hoop or whatever, and it's just there. This whole idea of using your imagination to draw something, and to get better at drawing, is maybe not going to be as important to a five-year-old. AI is part of their life now. And maybe these old ways of developing your intelligence and your creativity and your self-expression and your self-confidence are all just getting transferred into the AI. And the next generations of these tools are gonna get even better. Why would I go outside and take a photograph of a tree when I can just type a prompt to make a photograph of a tree?

Brendan:
Yeah. Well, people think I'm weird because I've never used Deliveroo or any of those food delivery apps. If we want take out, we go out and get it! <laughs> And a lot of people think, what are you doing? Why are you going in your car to get some food? And it's like, I just feel that's right for me. But I saw recently that someone from a design agency commented that their young interns don't know basics of using desktop computers, like downloading files. Because on the mobile devices they use, there's no such thing as files, you know, there's no desktop metaphor. It's kind of disappearing. So the interns were like struggling to attach a file to an email. So already a lot of those skills that we take for granted, that we've grown up with, are starting to erode.

Brendan Dawes, photo by Paul Dodd

Gary:
Yeah, and those things are not necessarily needed.

Brendan:
They're not. Well this is the point, you know, it's easy to sound grumpy about the past. "Well, in our day..." But I do see amazing content made. You know, the way younger people are using these tools is quite phenomenal. So I am optimistic. There are people making brilliant work. I think where it worries me more is on the consumer side.

Gary:
Do you mean with AI tools?

Brendan:
Well, no, the tools we all have access to, whether it's this bloody amazing TikTok video or whatever. They can really push a lot of those tools, where you go, "Wow, how have they even done that?" So I'm always amazed by what the younger generation can do, I'm not worried about that too much. It's kind of like that LCD Sound System song, you know, losing my edge? <laughs> All these young whipper snappers coming up behind me. But I think that people seeing the nuances is something I do worry about. You only have to look at what films are like now. You know, we judge things by value, unfortunately, by monetary value a lot of the time. And you know, the biggest films are things like the Marvel ones. Some of them are good, and they're entertaining, but they don't move me in the same way as watching a Lynne Ramsay film or a Jonathan Glaser film. Why isn't everyone talking about You Were Never Really Here, why haven't they all seen that film rather than Marvel, blah, blah, blah, you know? But then I think it's always been like that, hasn't it? Art, or the things that really move us sometimes, is not seen by the masses or doesn't appeal to the masses.

Gary:
Maybe, but I think we're now in a world where a lot of the choices about which films and shows get produced are based on algorithms and data on what people have watched before. So you get this constant narrowing down of what actually gets funded and made and shown. With the streaming platforms, if it hasn't done well on the platform before with audiences, whether it's the subject matter or particular actor or theme, then they won't take a chance on it. It's all about what's the lowest risk possible. And when I first started thinking about AI and these text-to-image generators, I was kind of thinking something similar. AI is only drawing from things it’s been trained on in that dataset, in that instance. It's only drawing from things that have already been made, that are already out there. And yes, it's re-combining things in interesting ways, but is it actually making something that doesn't exist yet? Or is it just mashing up stuff? But then I thought, well, isn't all creativity based on something else that already existed, and what we consider genius is simply a new interpretation or a new combination of existing things? Isn't this the way that human creativity has always been?

Brendan:
Yeah, that's it, because we make things based on things we've seen, and sometimes that might all get jumbled up and come out as something different. All art history is based on things that went before, you know, that were built on the shoulders of giants or whatever. So when I've seen people trying to defend AI, that's the argument they put forward, that this is how humans have always created. It's just that now we've made it automated and mechanized in a way. But even though there is an analogy between those two things, I think it's still different. I think AI is much broader than just, you know, Stable Diffusion and ChatGPT. There's a lot of amazing work going on that isn't talked about, like using AI to try to find cures for cancer. I guess that's all part of society's problem as well, that it's so knee jerk and so polarized that no one wants to drill down into detail because people just want simple truths. Right? Well, they're not even truths are they? They just want simple headlines. People love slogans, you know, and they don't want the detail. So you mostly see things in the news about AI ripping off artists and that kind of stuff. Which is a bit sad. I would encourage people to read more about it and drill down into the history of AI and read books about it. If they go deeper than the headlines, they'll find it's much more interesting.

Watch a short video about Brendan and his art, produced by Gazelli Art House: