The arrival of generative AI has put unprecedented pressure on a boundary that, until recently, was so natural we barely acknowledged it existed: the difference between making something and understanding it. Writers, designers and filmmakers have long known that early drafts are not expressions of clarity but tools for discovering it. Joan Didion famously said she writes entirely to find out what she’s thinking; architect Frank Gehry described his first sketches as “tentative gestures to feel my way into an idea”; cartoonist Lynda Barry wrote about drawing as a form of thinking rather than illustrating. Drafting has always been a process of revealing intention, not communicating it. But GenAI has changed the relationship between these stages in ways both promising and precarious.

When a model like GPT-5 can produce a dozen fluent paragraphs in ten seconds, and a system like Midjourney v7 can generate polished visuals before a concept has fully entered consciousness, it becomes easy to confuse expression with thought. The tools generate material so quickly that the act of drafting—the slow, uncertain effort through which meaning is usually shaped—can appear unnecessary. Yet this is an illusion. The speed of the system does not replace the slow work of the mind.

Generative models do not think. They assemble. They approximate. Their fluency does not represent introspection; it represents statistical prediction. A text that sounds complete may simply be coherent. An image that looks refined may simply reflect dominant aesthetics in its training data. What the model produces is not a conclusion—it is a suggestion. And confusing that suggestion for intention is one of the most significant creative risks of our era.

But rather than rejecting the technology, Copy Lab sees this as the moment that clarifies the true meaning of creative work. The sacred partnership between humans and GenAI depends on recognizing that AI accelerates drafting, but only humans can perform the thinking that gives drafting its purpose. The machine can generate variations, but only the creator can sense which variation aligns with a developing idea. The machine can supply form, but only the human can supply meaning. The machine can imitate the texture of thought, but only the human can perform the thinking.

This difference is not abstract; it is visible in every discipline. In literature, George Saunders talks about revising until a story reveals its emotional truth. In film, Barry Jenkins describes writing scenes to understand the psychology of his characters, not simply to fill a script. In design, Paula Scher often sketches dozens of intuitive versions before recognizing which one carries the right internal tension. Their processes demonstrate that draft-making is not merely production—it is cognition.

Generative AI offers something extraordinary: it takes on the laborious part of drafting, freeing creators to spend more time in the interpretive, strategic and emotional dimensions of the work. But this only works if creators remain responsible for the thinking itself. Without that separation, AI-generated drafts risk becoming masks—smooth surfaces that obscure the deeper work, the kind that cannot be automated.

There is also a more subtle danger: prematurely accepting the first AI-generated draft can lock an idea into place before it has had a chance to unfold. Anyone who has developed creative work knows that the first version is rarely the truest one. Often, it takes time—false starts, reframing, instinctive adjustments—before the underlying logic reveals itself. AI can produce endless versions, but these versions do not replace the sequence of interior shifts that allowed creators like Toni Morrison, Spike Jonze or Rei Kawakubo to refine their ideas through active engagement.

The invitation of GenAI, then, is not to move faster but to think deeper. To use the tools to widen the field of exploration, not narrow it prematurely. To allow the machine to provide drafts so the human can devote attention to interpretation. To ensure that clarity comes from within the creator, not from the surface polish of the model’s output.

This is why Copy Lab views the collaboration between humans and GenAI as sacred: because when the partnership is understood—and not confused—AI empowers creators to inhabit the creative process more fully, not less. It allows them to spend less time producing and more time perceiving. Less time arranging and more time discovering. Less time forcing ideas and more time hearing what they are trying to say.

The distinction between drafting and thinking has never been more vital. AI can handle the former. But the latter—messy, intuitive, emotional, intellectual—remains entirely human. And when those two forces work together, each fulfilling its true role, creativity becomes not only faster but richer, more intentional and more deeply alive.

/Carl-Axel Wahlström, Creative Director Copy Lab, 2025