The secret to taking a great photo, is to take a lot of bad photos.

Lots of people take bad photos and never a great one. It takes a "great" photographer of unique talent and vision to do that. So, why do all the great photographers say something along the lines of the above. Why do artists have basketfuls of crumpled up sketches? Why do novelists churn out one draft after another before settling on a final product with which they would be doomed forever to tinker were it not for the obstinacy of their editor and avarice of their publisher?

Everybody can create. Something. Some formats of creative expression are gated by a minimum of skill (painting, musical instruments) or access to tools and materials (sculpture, pottery) - but once they pass through a minimal threshold - creation is at anyone's finger tips.

So why do we insist that there is such a thing as creative genius that has somehow reached into a mysterious well of creation and under whose hands, the same clay arises into shapes completely unlike those of mere mortals? What makes Constable transcend the output of a local landscape artist, what puts Freddy Mercury above the singer in a cover band, Ansel Adams better than your uncle snapping holiday pics with a smartphone? When you look at them work, they go through all the same motions. If you were to watch them work, you'd probably not be able to perceive the difference. Why?

That's because the difference comes from choice. The act of creative genius is not in producing an output but in choosing from that output the one that will represent a vision, an ideal, a resonance - and one that will not only please them but that will speak to others - be recognised as beautiful, profound, somehow truer to life than life itself. That is what makes somebody a master of creation - perception.

This is becoming ever more important in the age of AI when the production of artefacts has become trivial and fully accessible to anyone. Anyone can now write a beautiful poem - and yes, AI can write beautiful poems. Anyone can generate a song that will delight and even move the listener. Anyone can produce an award-winning piece of digital art. And, now, anyone can produce a comic or a poster. Simply by typing in a few words describing what they want.

Where's the creativity in that? First, there is the intent. The idea that there is something to be expressed and the need to express it in a way that goes beyond the simple propositions of language. I don't just want to say "we're having a nice spring with many flowers outside and it makes me feel happy", I want to add another dimension to that expression that transcends the propositional. I want that dimension to resonate with my vision but also communicate it outside of myself.

To do this, I impose a constraint of form. I have infinite ways of expressing the fact, but I will want to reach for a particular subset of those forms that both limit and augment the expressive element. If I want to write a poem, compose a song or make a beautiful image that expresses the proposition about spring, I have to limit myself to the language of creation: Poetry, prose, music, painting, architecture - or even computer code. Constraints on form (and in many ways content) that will not only limit the exact shape of what I produce but will put demands on my ability to perceive it as belonging to its class and - in rare cases - be an excellent example of it.

I have to see that what I have created is good. In some forms, the creation is so expensive that I have to do the perception in my head or through a model but most of the time, I will try different permutations on the same theme until I find one that I can see is right. Some aspects of that rightness may be and more often than not are accidental. I may have stumbled on a rhyme, harmonic concept or a hue of a paint by accident. A fleeting whim after hours of fruitless attempts at purposeful choice. But not letting that accident go, knowing it for it is, that's what makes for a "creative genius".

That's why children are not truly creative. They are ultimate producers of form without filter. They generate many gems that make adult long for that freedom of imagination. But that's a mistake - children can produce flights of fancy but they lack the scope of perception to make a choice. The YouTube video series "Written by a Kid" illustrates it perfectly. In each episode, a child (6-7) tells a story which the adults than enact into a dramatic piece. The children's stories are full of imagination but are almost completely incoherent - the children have not only not mastered the narrative form, they are unable to perceive this lack. What they do is produce cute flights of imagination, but it is not creation only production - because it lacks perception.

This is not just about the arts, this applies to all sorts of creation of novel forms - inventions of new technologies, discovery of medicines, breakthroughs of science, theoretical transformations. They all depend on perception. Einstein's celebrated "annus mirabilis" was not a result of pure production of intellectual feats. His ability to produce the breakthroughs he did depended on him being able to perceive them as novel. His thought experiments did not just happened, he went on a systematic search of the idea space and did not stop until he found the ones that fit. But this was thanks to his ability to perceive the fit of those ideas compared to the ones that were not quite right.

This brings us back to the question of skill. Until now, people typically develop their perceptive ability through training in production. Most great artists went to art school, musicians study music, novelists go to classes of creative writing, and scientists, well, they study science. They then go through an apprentice period until they reach some sort of maturity. This does not just apply to the best in their field. Even an average artist or a scientist of little distinction have to go through this process. We always learn to perceive as part of the process of learning to produce.

In fact, developing the perceptive eye is often quite accidental. In most cases, education is not great at targeting it directly - it just happens. If we're lucky. Disciplines that rely on fluent performance - music, sport, language - have often developed techniques that work directly towards being able to recognise and perform the right forms - as does to some extent medicine. Art, photography and creative writing have critique sessions, physics and math have worked examples. But when the fluency of performance is less salient, education in those fields focuses much less on developing perception directly.

But, now as the actual performance becomes less important, perception needs to go to the forefront. We need to start developing the perceptive eye and rich mental models on which the eye can draw in a much more purposeful way. Some of this will require a level of productive competence - but this level will be much less. Just like a coach or a music teacher do not need to reach the same heights of performance as their best performing students, it will not be required that users of Large Language Models will be able to perform most of the actions as well. But they will need to have sophisticated enough perception to be able to judge the output.

What exactly does this perceptive ability look like? How do we develop it? The simple answer is we don't know yet. It is just too soon to tell. But we do have some analogues we can draw on. For example, we know that it is possible to be much better at reading or listening to something in a foreign language than it is to produce it. And within this skill, there are levels. At first, it's just recognising the unfamiliar sounds as belonging to that specific language. Second is identifying the general sense of what is being said through catching hold of the occasional word or phrase. Next comes being able to understand the nuances and finally being able to judge the grammaticality or even the aesthetic quality of the production. This takes time but requires relatively little production.

Even more simply, few people can write as well as they can read. Very few people can produce coherent poetry combining the imagery and language form required. But many can appreciate its nuance and even judge the quality of the form. At it highest level, analysing a poem is just as much a form of expertise as anything else. In the sciences, people are able to judge the quality of an invention or the relevance of discovery much better than they are able to invent or discover. This is why autodidacts are so often prone to follow outrageous ideas - they have not been socialised into a mode of perception that can filter them out.

To take advantage of the output generated by AI, it is necessary to be able to perceive it as relevant and appropriate to the task. And just like that of art criticism, this systematic ability to perceive is just as much as a skill that needs to be developed as the productive capacity. In fact, we have always been neglecting this foundational element of creation and can take this opportunity to develop it further. In this, just like in many other things, AI reveals shortcomings in many of our hidden assumptions about how we, as humans, go about the business of knowing.