Alex May

Executive Summary

Alex May's position on AI rests on a distinction. A programmer of more than forty years, he remains deeply sceptical of generative image models while embracing large language models as conceptual catalysts. His practice centres on creating what he calls digital mirrors: contemplative spaces that reflect viewers back to themselves and excavate feelings that do not yet have full names. Technology, for May, is a medium of intimacy rather than spectacle.

His most revealing AI integration happens before production begins. He maintains an elaborate Obsidian system of atomic notes drawn from readings and observations. When caught in intuitive tangles, he deploys LLMs to find unexpected connections, and over the past year this practice has genuinely shifted his artistic focus.

How I talk about my work today with you is very different than if we'd had this conversation a year ago.

He draws a firm line at generative imagery. The objection is philosophical rather than aesthetic. AI fills frames with information adequate enough to avoid distraction, but lacking the deliberate purpose every element in his work requires. His resistance to partnership framing is equally considered. Forty years of technical practice makes him cautious about substituting generated output for accumulated judgement. What he values is intelligent curation: AI that surfaces relevant material and helps thought develop without creating dependence.

The Zettelkasten Mind

May's most productive AI use happens in pre-production thinking. His Obsidian system captures atomic notes, fully formed contextual entries from readings, videos, and observations, which accumulate into a personal knowledge ecosystem spanning memory, time, technology, and human experience. When intuition stalls, he brings specific blockages to LLMs and asks them to connect things in different ways, the more esoteric the better.

The results have been transformative at the conceptual level. Researching his fascination with scarecrows, he asked AI about Kuebiko, a Japanese figure associated with agriculture and knowledge, often represented as a scarecrow-like presence who cannot move yet possesses wide knowledge of the world. AI also suggested possible folkloric connections between European scarecrows, medieval bird-scarers and the labour disruptions that followed the Black Death. The knowledge fed directly into his practice, providing what he describes as a menu of resonant material to choose from. He adds an essential caveat about checking whether the connections are actually real and not hallucinated.

He also deliberately avoids memory-enabled chat sessions. When ChatGPT's memory function began connecting robotics to every conversation he had, the over-connection became hindrance rather than help. He needs AI to address specific blockages, not impose continuity across unrelated problems. The tool should serve the thinking, not shape it.

The Rejection of Generative Imagery

May experimented with commercial image models and found them fundamentally incompatible with his practice. His objection turns on intentionality. AI fills the frame with information that passes the threshold of attention without rewarding it, generating backgrounds full enough to register as adequate without ever being deliberate. A generated cityscape is not interesting because the city itself is not intentional; it is something that is merely enough.

If you look at something like Hieronymus Bosch's Garden of Earthly Delights, everything is very intentional and tells its own story and has its own history. So much of that gets lost because [AI generated imagery] always ends up looking like some sort of Hollywood hero shot.

This distinguishes film set-dressing, acceptable as background filler, from artwork where every detail must carry deliberate purpose. AI's default aesthetic, the generically beautiful and the visually uncontested, fails this test structurally rather than incidentally. Even for prototyping, generative imagery does not satisfy him.

His deeper concern is what AI conceals from its users. Learning image construction, storytelling, lighting, and composition requires years of accumulated knowledge. AI does not encourage practitioners down those paths because it does not reveal its own shortcomings. The danger is premature satisfaction, capability enhancement mistaken for skill development. He is careful to clarify this is not a universal argument against AI tools but a specific warning about what gets bypassed when the bottleneck disappears before the understanding is built.

Childhood Computing and the Foundation of Control

May's relationship with technology traces to a formative revelation around age eight. Before computers, he loved cinema, made Super-8 animated films, rode BMX bikes and practised magic. His first experience of making something happen on screen through code had an immediate effect.

The computer became a world he could enter and understand from the inside during a childhood marked by frequent moves following his father's work. It was a place no one else around him fully understood, and therefore a space that felt his own. He still remembers the electric feeling of running that first program and seeing the output appear, describing it as hitting him like a train.

This origin story illuminates what follows. Technology serves as a medium for working with systems from the inside: a way to give form to the unnamed feelings, shared human states, and contemplative experiences his work attempts to reach.

I'm not trying to make cool stuff. I'm trying to connect to the [human experience] inside myself [through technology].

AI can assist the intellectual approach to that goal. It cannot substitute for the manual construction through which the work finds its form.

Digital Mirrors and the Private Performance of Code

May's aesthetic goal departs from demonstrative mastery. He is trying to create a small space, amid everything competing for attention. The work reflects viewers back to themselves so they find something to connect with or not. He seeks feelings that do not have full names, shared human states viewers recognise without being able to articulate.

Success occurs when a work becomes part of how someone sees the world.

That's the greatest honour in the world, that this has sort of become part of you, like a tool for you to think about the world with.

He positions himself as the instigator of an emotional space rather than a communicator of specific messages. His preference is for slow and meditative pieces that require genuine engagement, works that bring up feelings and connections in the viewer as a kind of private performance. This philosophy extends to his music, which he describes as a mist that hangs in the air. The music establishes mood; the visuals reinforce the state. Together they create something mesmeric.

Code, meanwhile, remains invisible to audiences, but it is not incidental. For May, programming is both material and method: a way of building the conditions in which the work can behave, unfold and affect the viewer. The technical elegance of the coding may be his private experience of making, while viewers encounter the manifestation of the idea. Since implementation serves emotional transmission rather than demonstrating capability, whether AI assists with implementation is less important than whether it changes the conditions through which the work acquires meaning.

Symbiosis, Dependence, and the Artisan Future

When the conversation turned to AI as creative partner or brain extension, May drew a distinction between assistance and dependence. His concern is not philosophical objection to AI's capabilities but practical worry about skill atrophy. If practitioners stop increasing their skills because AI can do the work, the monthly subscription becomes secured indefinitely while their own capability stagnates. Learning the skill yourself, by contrast, develops it in a personal way that does not leave you dependent on anything.

He acknowledges the counter-argument honestly. Most people do not value lifetime learning and self-challenge in the way he does. They want to make something, put it out, and move on. He recognises that not every artist wants or needs to build their own tools. But for May, the act of making the system is part of how artistic judgement develops. The skill is not only technical capacity; it is accumulated attention.

What he envisions as genuinely useful is not partnership but intelligent curation, AI that proactively surfaces relevant material from his interest domains without requiring constant querying, something closer to a thought of the day than a collaborator. Enhancement without dependence.

His broader concern is homogenisation. Once everyone can generate anything instantly, the question of differentiation becomes pressing.

If everyone can do it, what's the point of differentiation?

His hope is that, as generated imagery becomes more common, audiences will still seek out work that carries the trace of a particular human judgement: work shaped by sustained attention, accumulated skill and a distinct way of seeing. For May, that smaller but more engaged audience is enough.