Research Themes
Ten themes emerged from our conversations with 35 international artists working with AI. They map the patterns running through how practitioners actually use these systems, where current tools fall short, and what genuine creative partnership with AI could become. Together they form the spine of the forthcoming book.
How the Relationship Changes Over Time
A person's relationship with an AI tool is not fixed. It grows and shifts over months and years, through curiosity, frustration, small discoveries, and sometimes a decision to step back for a while. This theme looks at that long arc: how artists in music, visual art, design, film, and performance describe their history with these tools as something that develops rather than something they simply adopted. As the artist changes, the way they work with AI changes too, and the tools slowly reshape how they understand their own practice.
Tool, Assistant, or Partner?
Is AI a tool, an assistant, a collaborator, or something else? Artists answer this differently, and the disagreement matters more than any single answer. This theme looks at how they place AI along that range, from a machine that does what it is told to something closer to a creative partner, and at what changes depending on where they place it. The word an artist chooses says a great deal about how they see their own role, and it is one of the most contested questions in the field.
What the Body Knows
Much of creative skill lives in the body: in the hand that has drawn ten thousand lines, the ear trained over decades, the years of physical practice that become instinct. This theme asks what happens to that hard-won bodily knowledge when a machine can produce, in seconds, work that once took a trained hand to make. For a dancer, a musician, or a painter, the body is not a delivery system for ideas but the place the work comes from, and AI presses directly on that.
Wanting to Be Surprised
Artists across every field in this study want the same thing from AI: to be surprised in a way they can use. A tool that simply gives you what you asked for turns out to be less interesting than one that occasionally hands you something you did not expect. This theme looks at how artists chase that useful kind of surprise, how they tell a genuinely good accident from mere noise, and how the fast pace of AI can both spark that surprise and rush the slower thinking that creative work depends on.
Who Is Really Making the Work?
When an artist works with an AI system, who is making the work? The question is legal, ethical, and deeply personal all at once. This theme looks at how a sense of authorship holds up under AI, how artists protect the part of the work they feel is truly theirs while still letting the machine contribute, and how they draw the line between their own intention and what the machine generated. Underneath it sits a stubborn point: the machine produces, but it does not mean anything by what it produces, and the meaning is something the person brings.
A Tool That Forgets You
Most AI tools start from nothing every time you open them. They hold no memory of your taste, your past work, or the direction you have been heading in, so each session begins as if you had never met. This theme looks at what artists want from a tool that could actually remember: one that builds up a real sense of their work over time and grows with them. It also looks at the unease that comes with it, since a system that knows you that well raises its own questions about who holds that knowledge and what it is used for.
Working With AI Without Wearing Yourself Out
Working with AI can leave an artist energised or quietly hollowed out, and the difference is not always obvious at first. This theme looks at what keeps a creative life healthy over the long run when a machine sits in the middle of it. It covers the fear of losing skills you no longer practise, the pull of leaning on the tool too heavily, and the difficulty of staying yourself when the easy path is to let the system do more and more of the work.
Whose Values, Whose Cost?
Every AI creative tool carries choices made by someone else: what it was trained on, whose work it treats as normal, and who profits when you use it. This theme looks at the values built into these systems and the costs that the cheerful marketing tends to leave out. It covers the bias in what the models have learned, the handful of companies that own the infrastructure everyone depends on, the artists whose work was used without asking, and the real energy and money these tools consume.
The Work Becomes Choosing
Generative systems can produce images, tracks, drafts, and whole scenes quickly and at almost no cost. An artist can call up dozens or hundreds of usable versions in minutes, and once the options pile up, a large part of the work becomes deciding which ones are worth keeping. That deciding is a real creative act, and it rests on two things: knowing what is good, and knowing what is good for your own work in particular, which is what makes one artist's choices different from another's when everyone has the same tools. So when anyone can generate, the rarer skill is knowing what to keep, and questions of authorship, value, and who gets ahead increasingly turn on how well a person can choose.
What a Tool That Knew You Could Be
What would it feel like to work with an AI that genuinely understood your practice? This theme gathers the hopes and the harder-edged predictions artists offer when they imagine what these tools could become: instruments that respond to feeling, systems that hold years of your work in mind, partners that grow alongside you. It also looks at what stands in the way, which turns out to be less about what the technology can do and more about who owns it and who can afford it.
Interviews as Creative Encounters
I treat the interview as a creative encounter, acting as co-participant by bringing my own experience and pushing back. I ask sceptics to imagine AI in their work, and press enthusiasts on authorship, dependency, and lost agency. The conversation expands the practitioner's thinking in real time, and often mine too.
Reflexive layers run alongside the interviews: notes on my own AI practice, public essays that test how readers respond, and return interviews that track how practice and perception shift.
Close reading of each conversation surfaces patterns and contradictions across the corpus, with attention to bias and interpretive drift as themes develop.