How Generative AI Is Reshaping Creative Strategy

AI is not replacing creativity — it is expanding what is possible.

The Night the Brief Stopped Being a Wall

There is a particular kind of dread every creative strategist knows: the blank moodboard at eleven at night, a brief due at nine, and a brain that has already spent its best ideas on the last three campaigns. For years, that dread was simply the cost of doing the work. You pushed through it, or you didn’t, and either way the clock kept moving.

I remember the first time that changed for me. I was staring at a brief for a product I didn’t yet have a feeling for, and instead of waiting for inspiration to arrive on its own schedule, I opened a generative tool and asked it to argue with me — to give me ten directions I hadn’t considered, three of them deliberately strange. Within minutes, the blank page wasn’t blank anymore. It wasn’t finished, either. But it was moving. And in creative work, movement is often the hardest part to manufacture on demand.

That night reframed something for me. The tool hadn’t done the creative work. It had done the waiting — the part of the process where you sit with uncertainty until a shape emerges. What it handed back wasn’t a finished idea. It was raw material, fast enough and plentiful enough that I could finally afford to be picky.

Can a Machine Actually Be Creative?

It’s worth answering this honestly before going any further, because the article falls apart if it dodges the question.

In January 2026, a team led by Professor Karim Jerbi at the Université de Montréal — with deep-learning pioneer Yoshua Bengio among the co-authors — published the largest comparison to date between human and machine creativity, testing more than 100,000 people against leading generative models on standard divergent-thinking tasks, along with writing haikus, movie synopses, and short stories. The study, published in Scientific Reports, found that generative AI has reached a genuine milestone: it can now surpass average human creativity on these measures, though the most creative individual humans still clearly outperform even the best AI systems. That’s a real data point, not hype. It means the question “can AI be creative” no longer has an easy no as its answer.

But the study’s own authors were careful not to overclaim. As Jerbi put it in coverage of the research, even the strongest AI systems still fall short of what the most creative humans can produce — a meaningful, durable gap. A separate, equally important study out of the University of Houston’s Bauer College of Business, led by Professor Jinghui Hou, adds the texture that matters most for working strategists. Researchers there found that generative AI helps both novices and expert designers during ideation, the brainstorming stage — but at the next stage, once ideas need to be shaped into something finished, AI creates real friction for experts. As Hou explained, expert designers have years of training invested in how they materialize a piece of work, and revising what an AI produces at that stage can become genuinely burdensome. One report on the study put a number on it: in the lab experiment, expert designers who used AI spent 57 percent more time completing their work than peers who didn’t, even as the same tools kept helping less experienced people.

That distinction is the spine of everything else in this article. AI is at its best when the work is still wide open, and at its most limited when the work needs a trained hand to close it. Creative strategy has always understood the difference between opening a door and walking through it. Generative AI has simply made the door wider.

Where the Expansion Actually Happens

Ideation, at a volume no single mind can match. Where a brainstorm used to produce a handful of directions before the room’s energy ran out, a generative pass can produce dozens. Ari Kuschnir, founder of the studio m ss ng p eces, described this shift to Forbes as moving from the moment an idea sparks to a working execution in hours instead of weeks — while being clear that editing remains his own superpower, since AI still lacks the subtlety and human touch that stage demands.

Visual production, especially in the exploratory phase. Before a dollar is committed to a shoot or a set build, generative image tools let a team see a direction rendered, however roughly, and argue about it while it still costs nothing to change. A 2026 field guide published by the Association of Registered Graphic Designers (RGD) captures how this looks in practice: designers are using tools like Midjourney for conceptual, exploratory visuals — generally as a source of inspiration rather than finished art — and dedicated AI moodboard tools to externalize a feeling or theme in natural language before committing to detailed design work.

Iteration and localization at scale. A single campaign concept can now become dozens of tailored variants without a linear rebuild of the whole production pipeline each time. Coca-Cola’s Project Fizzion, co-developed with Adobe, is the clearest real-world proof of this: the system encodes a brand’s established design decisions — logos, type, imagery — into what the company calls a machine-readable “StyleID,” letting Coca-Cola teams and agency partners generate hundreds of localized campaign variations that automatically stay on-brand. The company reports content produced up to ten times faster without compromising brand integrity, quality, or originality — while stressing that designers remain fully in control of the process.

The Adoption Is Already Behind Us, Not Ahead

It’s worth pausing on how far this has already traveled, because many boardrooms still treat generative AI as an emerging bet rather than settled infrastructure. Adobe’s 2026 Creators’ Toolkit Report, developed with The Harris Poll and based on a survey of more than sixteen thousand creators across eight countries, found that 87 percent of creators using creative AI say it has accelerated the growth of their business or audience, and 75 percent describe it as integrated or essential to how they work.

But the same report resists the tidy, triumphant version of this story. Fifty-seven percent of creators told Adobe their AI outputs typically require moderate or extensive editing before they’re ready to share — a fact that quietly confirms the Houston research’s warning about AI’s limits at the finishing stage. And when Adobe asked what creators actually want as these tools grow more capable, the answer wasn’t more autonomy: 85 percent said the final creative decision should always rest with the human, regardless of how much AI or agentic AI is involved. Creators aren’t asking to be replaced by faster machines. They’re asking to keep the pen, even as the machine hands them more pages to choose from.

That instinct shows up in the market numbers, too. The generative AI creative-industries market is projected to grow from roughly $5.4 billion in 2026 to over $14 billion by 2030 — a trajectory that reflects mainstream adoption inside professional pipelines, not a passing experiment.

Prompting Is Not Strategy

If there’s one idea worth pulling out of this research and putting on a wall, it’s this one. Grace Liu, a creative director and brand strategist who has spent over a decade directing campaigns for global luxury and beauty brands, put it to Forbes plainly: prompting is a tactic. Direction is a discipline. It sounds like a soundbite, but the Houston research gives it real teeth — the reason AI creates more work for expert designers at the finishing stage isn’t that the tools are bad, it’s that turning a raw AI output into something that actually meets a professional standard requires exactly the kind of judgment a prompt can’t supply on its own.

This is exactly why the strategist’s job hasn’t shrunk. It has relocated. The scarce skill was never typing the right words into a box. It was always knowing what a brief needs to say and what a finished piece of work needs to look like before anyone — human or machine — starts executing against it.

The Part Nobody Gets to Skip

No honest account of this shift can leave out its costs, and any strategist who ignores them is building on sand.

Homogenization is not a fringe worry — it’s a documented effect. A widely cited study published in *Science Advances by Anil Doshi and Oliver Hauser found that when writers used AI-generated ideas as a starting point for short stories, the resulting stories were rated as more creative, better written, and more enjoyable — especially for less naturally creative writers. But the AI-assisted stories were also measurably more similar to each other than stories written by humans alone. The researchers describe this as a kind of social dilemma: individually, writers were better off; collectively, the pool of ideas got narrower. Follow-up research has found a more hopeful footnote worth holding onto, though — homogenization appears to trace back to how* generative AI is used in a given interaction, not to some unavoidable property of the technology itself, which means teams that deliberately push for divergence rather than defaulting to the first output can partly counteract the effect. That’s arguably the most actionable insight in this whole piece: sameness is a design choice, not a law of physics.

Unresolved questions around training data and rights. Generative tools are trained on large datasets, and how that training data was sourced remains a genuinely contested question across the industry — enough that a professional body like the RGD has updated its Code of Ethics specifically to address AI, and has been explicit that it sees “deep flaws” in how today’s systems handle bias, creator compensation, and transparency, and intends to keep pushing publicly for improvement. That’s a body representing working designers saying this, not just a critic on the sidelines — worth taking seriously rather than treating as a footnote.

Craft fatigue disguised as speed. Volume isn’t vision. The Houston research is the clearest evidence for why this matters in practice: more AI output at the finishing stage doesn’t automatically mean less work for an expert — sometimes it means more, because a fast first draft that misses the mark still has to be fixed by someone who knows what “right” looks like. It’s easy to mistake a fast pipeline for a good one, and a strategist’s job is increasingly to hold that line when a deadline tempts everyone to blur it.

Where This Leaves the Strategist

I don’t think generative AI has changed what creative strategy is for. It has changed what creative strategy has time for. The hours once spent waiting for a first draft, roughing out a dozen visual directions by hand, or building a placeholder set just to see if an idea works — those hours are increasingly available for something else: sharper thinking about audience, sharper argument about why an idea deserves to exist, sharper judgment about which of the fifty options a machine generated is actually the one worth building.

The research bears this out from every angle it’s asked. AI helps most when the work is still open and unformed, and needs a trained hand most right when it needs to close. It generates more, but “more” quietly risks becoming “more of the same” unless someone is actively steering against that drift. And across every source in this piece — from the researchers at Montreal and Houston to the creative directors interviewed by Forbes to the designers surveyed by their own professional association — the same idea keeps surfacing in different words: the tools have gotten faster, but taste, judgment, and the willingness to keep control of the final call haven’t gone anywhere. If anything, they’ve become more visible, precisely because when everyone can produce something quickly, the only differentiator left is whether what you produced was worth making at all.

The blank page, for me, doesn’t feel like a wall anymore. It feels like a door that opens faster than it used to. What’s on the other side of it is still entirely up to the person who walks through.

— Fatimah Romana