I was on the ferry this weekend, crossing Sydney Harbour. The sky was milky blue, and the sun shone as if it'd forgotten that winter was less than a day away.
One of the ferry crew stood a few meters away, the word 'CREW' fading on the back of his jacket. He watched as, up ahead, yachts raced across the harbour and then he disappeared inside the boat.
When he emerged, he carried a camera with a telescopic lens the length of my forearm. He hefted it up, took a few photos, and vanished back inside.
I started telling myself the story of his life—how he loved sailing and photography, how working Sydney Ferries was his dream job. Out here at sea, rolling gently between the Harbour Bridge and Opera House, he had time to capture the yachts he loved.
I tried to picture his photo: a yacht with its spinnaker slack, still searching for wind as it nudged forward in the midnight-coloured ocean.
And in that moment, I understood what AI can't do.
The Art of Waiting
That ferryman watched the world looking for an opportunity. He waited for a moment that stood out from all others—one that would only happen that one time, in that one place. When he sensed it approaching, he grabbed his camera and captured it as best he could.
That's creativity. We open ourselves to the world and, with practice, learn to distinguish the extraordinary hiding inside the mundane. Artists know that beyond talent or craft, you must teach yourself to genuinely see the world—to observe its details and idiosyncrasies, to notice what everyone else misses.
The photo my ferry photographer took would be unlike anything AI could create, because I couldn't begin to construct a prompt for it. What would I write? "Yachts racing on Sydney Harbour on a clear day with Port Jackson Bay in the background, wind gusty, spinnaker not yet catching, sky clear, ocean dark"?
That prompt captures none of what made the moment worth photographing.
Copies of Copies
Here's the fundamental problem: prompting AI doesn't allow for serendipity. LLMs adhere to your instructions, and our prompts can only extend from what we've already experienced or imagined. AI can't capture an unrepeatable moment in a particular place unless a person was there, observing it firsthand.
What we ask of AI is a facsimile of what we and others have lived. What AI delivers back is a copy of that facsimile—an artifact twice removed from the world and experiences that define our lives.
Art made with AI will always be a copy of a copy.
As David Hockney observed: "I can often tell when drawings are done from photographs, because you can tell what they miss out, what the camera misses out: usually weight and volume—there's a flatness to them." If photographs already flatten the world's dimensionality, imagine how much flatter AI's interpretations become, working as they do from datasets of those same photographs.
This matters because AI cannot feel the cold wind sweep over Namibian desert dunes. It doesn't understand how light changes as you climb to 6,000 metres in Peru. It has never heard the thunderous crashes of giant waves at Praia do Norte in Portugal when the surf is pumping.
As Sean tells Will in Good Will Hunting: you might know everything about Michelangelo from books; "But I'll bet you can't tell me what it smells like in the Sistine Chapel. You've never actually stood there and looked up at that beautiful ceiling."
AI's relationship to beauty is purely computational, colour wheels, symmetry, pattern recognition. It processes datasets of human experience but has none of its own.
Why This Still Matters
I'm not claiming AI art can't be moving or useful. But we must recognize its inherent limitations. When we mistake AI's sophisticated mimicry for genuine creativity, we risk forgetting what makes human art irreplaceable: the irreducible experience of being present in the world.
That ferry crewman didn't just take a photo; he witnessed something that will never happen exactly that way again and had the skill to recognize its worth. His image carries the weight of that singular moment, that specific light, that particular convergence of wind and water and human attention.
AI can generate a thousand yacht photos, each technically proficient and aesthetically pleasing. But none will carry the weight of having been there, of choosing that precise instant when the ordinary world revealed something extraordinary.
The world contains more than algorithms can capture, more than prompts can describe. And that's exactly why we still need people with cameras, waiting for the moment worth taking.
I think a lot of what you’re referring to is inspiration. Inspired comes from the Latin word ‘inspirare’ meaning ‘to breath into’. It’s this idea that moments like that crew member experienced, or the moment that inspired you to write about it, is like a breathing some kind of life or purpose into your soul , to move it to act in congruency with itself (the soul). I’d say it’s those moments where we’re closest to fulfilment.
One may also argue , in the world of gen ai, for now, we are the ‘beings’ breathing life into the machines. We create their inspiration - although if one believes the soul cannot be codified, then, that’s nullified.
Sorry for the tangent. Great post.