Exploring Artificial Intelligence and Art: Dave McKean Talks ‘Prompt’

From his collaborations with Neil Gaiman on The Sandman and Violent Cases to his seminal work painting 120 pages for Arkham Asylum: A Serious House on Serious Earth, Dave McKean is one of the industry’s true auteurs. He is out with a new book, Prompt: Conversations with AI where he explores artificial intelligence in the creative process. In an interview, McKean discusses his fascination with AI and whether a machine is capable of creating art.

Tom Smithyman: You joked that when you learned about AI image creation, you needed to retire or respond. How would you characterize your response via Prompt? What do you want people to take away from Prompt?

Dave McKean: On one level, I make books and projects about things that fascinate me, and the influence that AI will have on almost all aspects of our lives in the near future – and on what I do now – is interesting. It’s a huge complex subject, and it will enable all kinds of interesting new things, and there will also be collateral damage, like every other new technology that’s come along in the past. There have been many new things that have come along while I’ve been working professionally as a creative person, and many I’ve been happy to ignore – they’re just not interesting to me. But AI is not a new medium, or just a new tool (although it is that as well of course), it’s a fundamental resetting of our understanding of what creativity is.

That’s too big to ignore really, so my response was an adrenaline-fueled brain dump, where I tried to set down what I thought were the implications of this new tech, the good and the bad, and to at least try and start a conversation that looked ahead to the possible pitfalls, before it’s too late to do anything about them. And yes, I realize it’s probably too late already, but I still think it’s worth thinking in broader and deeper terms than just how superficially dazzling all this new stuff is. I hope people take away a little food for thought. I think we tend to just accept new tech as some sort of given, just the way things are. Technology is not prescriptive, we can and should feed back into the tech loop to try and improve the advantages and mitigate against the disadvantages.

Smithyman: What was your process for writing these stories? Was there any back and forth with the AI?

McKean: I quickly decided I wanted to make a book, and that it should be in comics form, simply because I think most clearly in sequences of images and narrative. I felt I needed to set up a couple of procedural ideas to teach myself how best to use the AI bot, to see what it could do and to show its power to my audience, to get a feel for how the algorithms work – what their strengths and weaknesses were.

The first story idea was to take a part of the epic of Gilgamesh – our oldest known human story – carved in cuneiform on stone by anonymous craftsmen, and feed it back into AI via our most current translation, to see if anything of the narrative would survive this transfer across 4,000 years and the turbulence of internet imagery. The second piece was simply an attempt to see how AI sees us through one newspaper headline per day. For the last story I took all my thoughts and observations gathered during those first two exercises and structured them through my usual problem-solving morning walk, a conversation with myself and AI, to see where those thoughts and questions would take me. The whole process from first inkling of an idea to a final finished pdf was 12 days.

Smithyman: Were there any AI responses that surprised you? Did they all make sense, or was anything lost in translation?

McKean: A whole range. Some were surprising because it’s sampling from the entirety of online imagery, so, like an infinite amount of monkeys tapping away on their typewriters, eventually something’s going to make a kind of sense. Some of it was bland. I had to abandon one of the headlines from part two, about the Texas school shooting, because the AI just kept sampling photos of grieving children and parents, and I felt it was immoral to co-opt their grief for my academic experiment. I was left with an uncanny feeling of talking to an alien intelligence, wondering where these images came from. All of it had a soulless, cold quality. Every time an image popped out of the AI that seemed like it was more than just a surreal mushing together of online imagery, I realized I was finding that significance there, it was not intended by the bot.

Smithyman: You put the book together very quickly – in only 12 days. I imagine your usual process takes considerably longer. What are the tradeoffs as you think about quality vs. expediency?

McKean: My argument is that they are fundamentally different things. I could not make the images that I try and create for my usual comics using AI. Maybe in the years to come it will get so good at mimicking human creativity that the difference will be too small to notice, but at the moment AI cannot make images I consider to be useful to tell the kind of stories I want to tell in the way I want to tell them. AI can create a certain kind of image, and I’ve already found it useful for standalone imagery for album covers or illustrations. I prefer using AI when I’m using it as a raw material generator and then taking that stuff and creating something more considered out of it.

Smithyman: A lot of artists are looking down their very human noses at AI art. But isn’t there a point in the future where you think AI will be the norm, at least for some basic writing or artistic tasks, or will human creativity always be preferred and dominant?

McKean: This is a huge topic as there are many different aspects of creativity, many different uses for it. There are obvious commercial aspects of image creation that AI can do perfectly well now, so there is no necessity to hire illustrators or designers for those jobs. But there are still many areas of creative activity where AI has nothing to offer, and then a huge spectrum of relationships in between, where AI is a generator of interesting stuff to be used, or a blue-sky generator of random conjunctions that may provoke human ideas, or who knows what in the future.

The big difference for me between AI-stuff (and I’m not giving it the ‘A’ word yet) and human creativity is that art is not just about getting to an end result. It’s about the process, the learning, the work, the testing of yourself, the flow of the conversation, the act of growing up. For me the art is about the doing of it, not the end result. This has nothing to do with typing a word into a computer and pressing go, just like playing the piano is not going into a record store and asking the sales assistant for a cd of piano music. I also hugely value the act of creativity as the defining center of my life, and as a powerful aid to mental wellbeing for many.

Smithyman: After your experience with Prompt, do you believe that art and its creation are uniquely human concepts?

McKean: Yes I do. That’s not to say humans won’t find ways of incorporating AI into their work. Of course they will, they are bound to. AI will be an evolutionary change to humankind in this century, it will affect all of us. I do still believe though, that the thing that makes a drawing “art,” for me at least, remains the act of seeing through the eyes, and the life experience, of the person who made the drawing.

Smithyman: On a different topic, you are known for – among many other things – your work on Sandman. What was it like seeing so much of your Sandman work translated onto the screen from Netflix?

McKean: I haven’t seen it.

Smithyman: Thanks very much for your time and your thoughtful answers.

Prompt: Conversations with AI is available now at davemckean.com.