Is Creativity About Creating or Choosing?

In 1915 Marcel Duchamp bought a snow shovel, hung it in his studio and called it In Advance of the Broken Arm.

He had not made the shovel. He had not developed a new material or spent months working on it. What he did was far simpler: he chose it.

Duchamp's "readymades" changed the familiar idea of what an artist does. As MoMA's notes on the work point out, Duchamp was showing that the creative act of an artist could come not only from manual skill but also from the act of choosing. See the work at MoMA

A century later the same question is back in a different form.

GenAI can hand us dozens of headlines, campaign ideas, visual concepts, customer dialogues or product scenarios in seconds.

The question is no longer only "Can we produce this?"

It is: which of these should we choose?

The blank page problem gives way to the full table problem

For a long time, creative work started from scarcity.

An advertising campaign needed an idea. A product needed a name. A new customer experience had to be designed. There was a blank page in front of you, and filling it took time.

GenAI changed that equation.

Today a marketing team can look at a hundred headlines instead of ten. A designer can try dozens of alternatives instead of a single visual direction. A product team can build different user scenarios far faster.

That is a significant gain in productivity. But more alternatives do not automatically mean more creativity.

In a study published in Science Advances in 2024, stories written by people who took ideas from generative AI were rated as more creative and better written. Yet those same stories also turned out to be more similar to one another. In other words, AI can improve the work an individual produces while reducing overall diversity. Read the Science Advances study

This finding reminds us of something important:

As production capacity grows, the value of creativity shifts from producing to choosing.

When everyone can produce a hundred options, producing a hundred options is no longer an advantage in itself.

The advantage is being able to see the one that stands apart from the other ninety-nine.

The human's job does not end when AI produces

There is a misconception we come across often. As AI takes over production, people assume humans will step back from the creative process.

The picture we see at CBOT is different.

As GenAI expands production capacity, the weight of human judgment grows with it. Every output brings new questions behind it:

Does this idea fit the brand? Does it repeat what we have already done? Is it grounded in a real customer insight? Could it get attention today and harm the brand six months from now? It looks creative, but does it actually work?

These are not questions for the model alone. They are questions about the organization's experience, strategy, culture and goals.

A large-scale PNAS Nexus study of text-to-image AI use shows the same double effect. Looking at more than four million artworks, the researchers found that AI use increased creative productivity by 25 percent, while average content novelty declined. Read the PNAS Nexus study

So we are producing more.

But producing more does not always mean we are thinking more differently.

The new creative skill: good editing

A film director does not use every scene that was shot.

A photographer does not exhibit every frame.

A writer does not keep every sentence in the book.

For years, a large part of creative work has been as much about deleting, filtering, ordering and letting go as about producing.

GenAI simply makes that truth more visible.

Until now we have mostly defined creative talent as "coming up with good ideas". In the years ahead we will need to add another skill: recognizing a good idea.

The two are not the same.

A model can produce a hundred slogans. Understanding which one would break the language the brand has built over years, which one sounds too much like a competitor, or which one will genuinely land with customers is a different skill.

That is why writing a prompt is the beginning of the process. Not the end.

The real value emerges in the selection process that starts in front of the options the model has produced.

The new bottleneck for companies may not be ideas

This shift is not limited to creative teams.

In large organizations using GenAI we see a similar transformation in product development, customer experience, innovation and operations design.

The problem is less and less "we cannot come up with alternatives".

The real problem is this:

With this many alternatives, which one do we invest in?

This is why we believe organizations need to build four capabilities together in how they use GenAI:

1. Define the selection criteria. Decide on measures such as brand, customer, performance and business goals instead of "we liked it".
2. Build a filtering mechanism. Quickly set aside outputs that are repetitive, average or out of step with how the organization works.
3. Put human judgment in the right place. Position people not as an approval step that checks every output, but as the owners of high-value decisions.
4. Feed good choices back into the system. Carry the discernment the organization develops over time into prompts, knowledge sources, agent behavior and evaluation systems.

The fourth point matters most.

An organization's real competitive advantage may not be the foundation model it uses. Many companies can access the same models.

The difference may come from how well the organization can teach AI what it considers good.

Can taste scale?

In art this is sometimes called taste, sometimes curation, sometimes editing.

In business we tend to call it decision quality.

GenAI is creating an interesting period in which these two worlds move closer together.

As the cost of production falls, the value of good selection rises. For a CMO, knowing which creative will not run becomes as important as knowing which one will be produced. For a product manager, seeing which feature should not be built gains value. For a customer experience team, choosing the on-brand option among the dozens of ways AI can phrase a conversation becomes a new design discipline.

That is why at CBOT we do not see GenAI systems only as "systems that produce".

We see their real potential in systems that also improve the quality of an organization's decisions.

AI can widen the set of alternatives, spot similarities, surface past performance and compare different scenarios.

But at some point someone still needs to look at the table and say:

"Let's go with this one."

Conclusion: a new symbol for creativity

For many years the symbol of creativity was the blank page.

There was nothing there. You had to create something.

With GenAI that image is changing.

What sits in front of us now is not a blank page but a table full of options. Texts, visuals, ideas, scenarios, alternatives…

In this new world creativity does not disappear. It moves.

Producing still matters. But in a period when producing has become cheaper and faster, looking, distinguishing, filtering, combining and choosing become more valuable.

Perhaps the creative person of the GenAI era will not be the one who produces the most ideas.

It will be the one who can see which idea is different from the rest.

That is why the question Duchamp asked more than a hundred years ago feels surprisingly current:

Does creating always require producing?

GenAI may be giving us a new answer.

Sometimes creativity is less about saying "make this" and more about being able to say "this is it".