For most of modern creative history, the bottleneck was execution. To make the thing, you needed the time, training, money, software, craft, and often a team backing you up.
In the last few years, that paradigm has faded into irrelevance thanks to the power and widespread availability of generative AI that can draft copy, generate concepts, compose visuals, localise content and produce variations at a speed and cost that would have seemed implausible only a few years ago.
Stanford’s 2026 AI Index reported that generative AI had reached 53% population adoption within roughly three years (faster than both the computer and the internet) and that the cost of inference for systems at roughly GPT-3.5 level had fallen more than 280-fold between late 2022 and late 2024. The World Economic Forum, meanwhile, says AI-powered tools are drastically lowering technical and accessibility barriers across media, entertainment and sport.
That's why the debate about human creativity has become more interesting, not less. The question is no longer whether machines can produce content that looks polished, plausible or commercially usable. They clearly can. The harder question is what becomes more valuable when polished, plausible and usable are no longer scarce.
Paul Aitkenhead, Head of Brand PR at Gamma, gets to the heart of it:
“One of the biggest misconceptions about AI is that it replaces creativity. I do not think that is true. What AI really does is commoditise average execution. It can help structure ideas, summarise research and generate first drafts quickly, but it still struggles with the things that actually make brands memorable and differentiated.”
That distinction matters. It moves the conversation away from the melodrama of “AI versus creativity” and towards the real economic shift already underway. Average execution is becoming cheaper. Abundant output is becoming less valuable. Discernment is becoming more valuable.
In the UK, that shift is no longer just a philosophical argument; it is a policy and market issue. In March 2026, the House of Lords Communications and Digital Committee warned that the creative industries face a “clear and present danger” from widespread unlicensed use of protected works and from weak transparency around AI training data.
The government’s own March 2026 report later confirmed that its originally preferred broad copyright exception with opt-out was no longer its preferred way forward after strong opposition from creators and the wider creative industries. Those concerns are not abstract: the same Lords report notes that the UK creative industries contributed £124 billion in GVA in 2023 and employed 2.4 million people.
In other words, the age of AI is not making human creativity irrelevant. It is making it easier to see where the real value was all along.
AI Has Made Creativity Easier, Not More Valuable
Rowdy Studio
If you want one sentence that captures the current moment, it is probably this: AI has made creative production easier, but it has not made creativity itself more valuable.
The distinction matters because ease and value are not the same thing. Lower barriers are good for access, experimentation and productivity. But lower barriers also compress the price of whatever sits on the other side of them.
The World Economic Forum’s 2025 report on AI in media, entertainment and sport is explicit on this point: AI-powered tools are expanding content development by lowering technical barriers, making content production easier and allowing more people to participate in creation. That is significant progress. It is also a direct challenge to anyone whose value proposition depended mainly on being able to make competent material faster than everybody else.
The research already reflects that tension. In a 2025 experimental study of creative writing, participants who used ChatGPT produced work judged to be more creative on several measures, with less inaccuracy and ambiguity and in less time. But they also reported that the task felt less effortful, less intellectually demanding, less enjoyable and less valuable. The authors conclude that AI assistance can improve output while simultaneously diminishing parts of the experience and meaning of the creative act itself. That is a revealing split. The performance went up; the felt ownership of the work went down.
A similar pattern appears in the visual arts. A 2024 PNAS Nexus study using a dataset of more than 4 million artworks from over 50,000 users found that text-to-image AI significantly increased artists’ productivity and improved peer evaluation, but it also reduced average novelty over time. The same paper argues that the people who gained most from AI were not those who simply pressed the button hardest; they were the ones who could explore more novel ideas and filter outputs for coherence and meaning. In other words, AI improved throughput, but the advantage still went to humans with better judgement.
That is why Aitkenhead’s framing is so useful. The real disruption is not that AI has suddenly invented creativity from nowhere. It is that it has removed friction from a huge amount of production work that used to carry quite a lot of value simply because it required skill, time and repetition. As he puts it:
“The most valuable skill in the future will not simply be producing more content faster. It will be understanding what is actually worth saying in the first place. Human judgement, positioning, taste, emotional nuance and originality become significantly more important when everyone has access to the same generation tools.”
This is the key commercial reality. When AI makes output abundant, output alone loses its scarcity premium. Speed becomes table stakes. First drafts become cheaper. Variations become infinite. But abundance does not create meaning. It creates a filtering problem.
That is what Becky McOwen-Banks, Founder at Plain:AI, is really pointing to when she says the disruption is not just technological but psychological:
“Let me tell you what I've noticed in every AI workshop I've run in the last two years… The moment I ask, ‘what's your creative superpower?’ the room goes quiet. Not because people don't know. Because they've started to doubt whether it still counts. That's the real disruption. Not the tools. The doubt.”
The doubt is understandable, but it is misdirected. If AI can do more of the making, the premium shifts upstream. The most valuable creatives in the age of AI will not be the people who can churn out the most competent material in the shortest time. They will be the people who know when the competent thing is the wrong thing, when the polished option is the bland option, and when the brief needs to be challenged before the first output is generated. That is not less valuable work. It is more valuable work.
Human Creativity Is No Longer Defined by Production
Denis Giuffrè
One of the most confusing aspects of the AI debate is that people keep using “creativity” to describe very different things. Sometimes they mean fluency: the ability to generate lots of ideas. Sometimes they mean originality. Sometimes they mean execution. Sometimes they mean emotional resonance. Sometimes they mean cultural significance. AI is changing all of those categories, but not in the same way.
On a narrow set of divergent-thinking tasks, large language models now perform extremely well. A 2024 Scientific Reports paper found that GPT-4 outperformed human respondents on several divergent-thinking measures, while another Scientific Reports study from 2023 found that AI chatbots outperformed humans on average in an alternate uses task. But that same 2023 study made an equally important point: the best human ideas still matched or exceeded the chatbots. T
he AI advantage was in the average. The human advantage remained at the top end. The 2024 authors also caution that such tests capture only one slice of creativity and do not necessarily establish superiority in usefulness, appropriateness or broader creative judgement.
That should change how the creative industries talk about the threat. The argument is not really that AI has become “more creative” in the full human sense. It is that it has become frighteningly good at reliably generating above-average material in tightly scoped formats. That is enough to put enormous pressure on committees, agencies, junior production roles, content farms and any workflow built around “good enough” deliverables.
The strongest peer-reviewed evidence on this comes from Anil Doshi and Oliver Hauser’s 2024 Science Advances paper. In an experiment on short-story writing, access to generative AI ideas made stories more creative, better written and more enjoyable, especially for less creative writers.
But those AI-assisted stories also became more similar to one another. The authors describe this as a social dilemma: individuals benefit, while collective novelty narrows. That is a remarkably good description of what many agencies and in-house teams are starting to feel. The floor rises. The middle gets better. The work converges.
Beto Nahmad, Executive Creative Director at VCCP, frames that shift with real clarity:
“For most of modern history, creativity was constrained by execution… Today, that constraint is disappearing. Artificial intelligence can generate images in seconds, write convincing copy on demand, compose music, edit video, and produce an endless stream of content… Creation itself is becoming abundant. What was once scarce is now available at the click of a button. And yet, as abundance grows, something else becomes more valuable. Taste.”
That is exactly right. Human creativity is no longer defined primarily by the ability to produce a polished object. Production is being democratised and automated. The more defensible part of creativity is shifting towards framing, choosing, discarding, sequencing and meaning-making.
McOwen-Banks frames that shift with real clarity:
“Faster was never the ceiling. The industry's default response to AI has been to reach for efficiency. More output. Faster turnaround. Lower production cost. And yes, AI can do all of that. But when did faster become the ambition? The creative leaders I trust most aren't asking ‘how do I use AI to go faster?’ They're asking ‘how do I use AI to get sharper?’”
That distinction between faster and sharper maps neatly onto the research. AI is very good at expanding option space. It is much less good at telling you which option matters, or whether the question itself is wrong. A 2025 systematic review of human-AI co-creativity found that high user control leads to greater satisfaction, trust and ownership over creative outcomes, and that there are still important gaps in AI support for early creative phases such as problem clarification. In other words, the fuzzier, more strategic, more interpretive parts of the process remain deeply human terrain.
So yes, AI can produce. Increasingly, it can produce impressively. But human creativity is moving away from mere production and towards something more editorial, more strategic and more selective. Or, as Nahmad puts it, “Great work comes from selection.” That sounds almost old-fashioned until you realise it is also the most future-proof definition of creative value available.
Why Human Taste Is Emerging as a Competitive Advantage
John Cooper
If AI is exceptional at generating plausible options, then taste becomes the scarce skill that turns possibility into value.
This is not just a romantic defence of artistry. It is what the evidence increasingly suggests. A 2026 PNAS Nexus paper comparing humans and many different LLMs across standard creativity tasks found that LLM responses were far more similar to one another than human responses were to each other. The authors warn that if widely used LLMs behave similarly, they may drive users toward a narrower set of “creative” outputs regardless of which model they choose.
A separate 2025 comment in Nature Human Behaviour reached a similar conclusion on brainstorming: ChatGPT can increase the average creativity of ideas while reducing the diversity of ideas in the pool. And the 2024 Science Advances story-writing experiment found the same basic trade-off. Better individual outputs, narrower collective novelty.
That is why McOwen-Banks’ line feels so potent:
“The averages are getting better. That's the problem.”
It is the problem because average no longer looks obviously average. It looks polished. It looks presentable. It often looks professional. But it also starts to sound, feel and move like everything else.
Simon Manchipp, Founder at SomeOne, says it more bluntly:
“Facts. The machine has the brush, but it doesn’t have the eye. AI can give you a thousand options, but it can’t tell you which one will make someone cry. Taste is the ultimate human moat. This is not news, taste has ALWAYS carried a premium. From choosing a guy with one name to paint the chapel ceiling to seeking out Coco for the dress.”
That last line is witty, but the argument is serious. Taste has always sat above execution in the hierarchy of cultural value. The difference now is that execution is being industrialised at such speed that taste is becoming newly visible as a commercial differentiator rather than a vaguely arty flourish.
Shagorika Heryani, Founder at Athina, gets at the same truth from another angle:
“AI optimises for flawless execution, but flawless is boring and everywhere. Human taste is lived experience. It is context, nuance, and the courage to choose the illogical idea that somehow lands perfectly. We connect through the flaw, the vulnerability, the things a AI model is literally programmed to smooth out. True tastemakers have always known this. They don't make things people like. They make people feel things they didn't know they were waiting to feel.”
That phrase “lived experience” matters because it captures something the current generation of models does not truly possess. They can simulate registers, references and patterns. They can average style. They can infer likely next moves. But they do not bring biographical risk, embodied memory, social instinct, embarrassment, grief, desire, obsession or contradiction to the work in the way people do. They do not know what a brand sounds like after a crisis, what a joke means in one community but not another, or why a safe sentence can still be the wrong sentence in a live cultural moment.
Nahmad’s definition of taste is especially useful here because it strips out the pseudo-luxury connotations and grounds it in judgement:
“Not taste as a matter of personal preference, but taste as judgement. The ability to recognize what matters. The instinct to know what deserves attention and what does not. The sensitivity to identify the difference between something that is merely competent and something that feels alive.”
That is the job now. In a market full of competent outputs, the creative premium moves to the people who can tell the difference between what is merely correct and what is actually alive.
The World Economic Forum’s 2025 New Economy Skills report makes a strikingly similar point in more formal language. It argues that tasks tied to empathy, creativity, leadership and curiosity have only about 13% potential for AI transformation because they depend on human judgement, context and lived experience, before concluding that “in the age of artificial intelligence, the true competitive edge is being human.” That is almost the policy version of Manchipp’s “human moat.”
So when people ask whether taste can become a competitive advantage for brands and creatives, the answer is not merely yes. It already is. The more abundant generation becomes, the more valuable the edit becomes.
The Skills AI Can't Replace Are Becoming More Valuable
Gidudu Emmanuel
The awkward thing about the phrase skills AI can’t replace is that it can sound like a defensive wish rather than a serious argument. The better way to frame it is this: which capabilities remain hardest to automate, easiest to undervalue, and most commercially important once routine production gets cheaper?
The World Economic Forum offers one answer. In the Future of Jobs Report 2025, employers rank AI and big data among the fastest-growing skills, but they also expect creative thinking, resilience, flexibility, curiosity and lifelong learning to rise in importance through 2030. The same report says that if the global workforce were represented by 100 people, 59 would need training by 2030, which is another way of saying the market is already being re-priced around new mixtures of technical fluency and human judgement.
That fits neatly with Aitkenhead’s view:
“We are entering a period where average content becomes abundant and therefore less valuable. In that environment, the brands and individuals who stand out will be the ones who understand their audience deeply enough to identify meaningful insights, tell relevant stories and create ideas that genuinely resonate. Ironically, I think AI will make human perspective more commercially valuable than ever. Because when everyone can produce content, discernment becomes the real differentiator.”
Discernment is the umbrella term. Underneath it sit several specific abilities that matter even more in an AI-saturated market: framing the problem correctly, spotting the insight nobody has articulated yet, sensing when a message is emotionally wrong even if technically polished, understanding timing, bringing a distinctive point of view, and knowing when not to make the obvious thing.
There is good evidence that audiences still respond differently to those human signals. A 2025 Journal of Business Research paper found that when consumers believed emotional marketing communications were written by AI rather than a human, positive word of mouth and loyalty declined. Crucially, the effect was much weaker for factual communications and weaker still when the AI was perceived as editing rather than authoring. That is a practical lesson for brands: audiences are not equally sensitive to AI across all communication contexts. Emotional and relational messages seem to carry a stronger expectation of human authenticity.
The same pattern appears in art. A 2023 study published via the University of Cambridge repository found that people judged identical artworks more positively when they were labelled “human-created” rather than “AI-created”, and that perceived story, effort and human engagement helped explain the preference. In other words, part of what people value is not just the final surface of the work but the human process behind it.
That matters enormously for anyone building a brand. Because the things AI struggles with are often the things that create lasting distinction: voice, vulnerability, emotional precision, cultural reading, social trust and narrative coherence over time. A one-off output can be generated. A body of meaning still has to be authored.
McOwen-Banks puts it in more operational terms:
“You cannot prompt for taste. You can approximate it, if the training data is good enough. But the model doesn't know your client, your audience, the specific cultural moment you're trying to meet. It doesn't carry twenty-five years of knowing what lands and what doesn't. It doesn't have the scar tissue. That's yours.”
There is the real value of human creativity in the age of AI: not that people are always more inventive on demand than machines, but that people bring context-rich judgement to ambiguity. They can read the room. They can absorb contradiction. They can understand what a brand is trying to become, not just what similar brands have already done.
That is why the skills AI can’t replace easily are becoming more valuable economically as well as creatively. They are not decorative human extras sitting on top of machine efficiency. They are what stop efficiency from flattening everything into the same answer.
Why Great Ideas Still Need Human Judgment
Oleg Buyevsky
If AI has a superpower, it is generation. If humans retain a decisive advantage, it is judgement.
That sounds obvious until you realise how rarely the creative industries have had to make that distinction explicit. For years, good work could smuggle its thinking inside the final artefact. A great campaign, identity or script did not have to explain the invisible sequence of decisions that made it good. In the age of AI, that invisibility becomes a problem, because clients can mistake generated abundance for solved thinking.
A useful concept here comes from educational research rather than advertising. In a 2024 paper on generative AI and assessment, Margaret Bearman and colleagues argue for developing “evaluative judgement”: the ability to identify and calibrate quality in outputs and processes involving generative AI. They describe this as a distinctly human capability that matters more, not less, in an AI-rich environment. The idea translates neatly to creative work. The premium is not merely on producing options. It is on recognising which option has merit, which one is derivative, which one is emotionally hollow, and which one deserves another three hours of discomfort before it is ready.
That is why Nahmad’s formulation feels so complete:
“AI is extraordinarily effective at generating possibilities. It can offer hundreds of concepts, thousands of variations, and an almost infinite number of creative directions. But possibility alone has never been the source of great work. Great work comes from selection.”
Research on human-AI co-creativity supports that emphasis on human control. A 2025 systematic review of 62 co-creative systems found that higher user control is associated with greater satisfaction, trust and ownership over creative outcomes, while also noting that AI support remains limited in early phases such as problem clarification. That is a critical detail. AI can be very useful once the frame exists. It is often less useful in deciding what the frame should be.
McOwen-Banks has a name for the commercial version of this: the Human Premium.
“If AI compresses the value of speed, volume and execution — and it is already doing exactly that — then what remains is the quality of human thinking applied to a problem. The judgement. The point of view. The creative intelligence that shapes what the tools produce rather than just operating them.”
She then adds the harder bit, which most creative businesses still have not fully internalised:
“But here's the uncomfortable part. You can't charge for what you don't make visible… The Human Premium only has value when it's legible.”
That is a profound business point. If your best contribution is the invisible sequence of rejections, reframes, edits and strategic interventions that turned generic material into distinctive work, then you increasingly have to help clients see that sequence. McOwen-Banks calls this “Visible Judgement”: showing where human experience overrode the model, where taste intervened, where context changed the direction, where the strong option was not the statistically likely option.
There is research backing this up in adjacent domains too. In the PNAS Nexus art study, the artists who benefited most from text-to-image AI were the ones able to explore novel ideas and filter outputs for coherence. The authors explicitly argue that ideation and filtering remain pivotal human skills in AI-assisted workflows. The machine expands the search space; the person decides what deserves to survive it.
This is why human creativity becomes more, not less, valuable as the tools improve. The better AI gets at making, the more expensive good judgement becomes. Cheap abundance makes discernment rarer. And rarity, as ever, is where value accumulates.
How Agencies, Brands and Creatives Can Develop Better Taste
Vinyl Impression
The practical response to all this is not to romanticise the human and reject the tool. It is to redesign what creative excellence looks like.
First, creatives need to stop treating AI fluency and human originality as opposites
The strongest position is usually a hybrid one: use AI to remove dead time, accelerate lateral exploration and pressure-test routes, but keep humans firmly responsible for framing, selecting and final meaning. The co-creativity research points in the same direction: user control, not passive dependence, is what improves trust and ownership. The WEF also argues that responsible adoption depends on a human-centric approach, workforce development and strategic leadership rather than technology alone.
Second, agencies and brands need to become much more deliberate about how they train taste
For years, junior development often happened through execution: make more decks, write more copy, produce more formats, learn by repetition. Some of that still matters. But if average production is increasingly commoditised, then people need earlier exposure to critique, judgement, cultural research, narrative logic, consumer psychology, positioning and the practice of defending why one route matters more than another. That is not an anti-AI curriculum. It is a post-abundance curriculum, and it aligns with the broader labour-market shift towards creative thinking, curiosity and human-centred skills.
Third, brands need to be more careful about where they let AI front the relationship
Consumers react more negatively when emotional messages are perceived as AI-authored, whereas factual messages are less sensitive and AI editing is less problematic than AI origination. The lesson is simple: if the communication is functional, AI assistance may be fine. If it is trying to convey care, empathy, gratitude, pride or inspiration, a visibly human hand matters more. That should influence not only copy workflows but disclosure, approval processes and brand governance.
Fourth, creatives need to get less embarrassed about saying that taste has commercial value
McOwen-Banks is right that “speed is a commodity. Taste is not.” Simon Manchipp is right that “taste is the ultimate human moat.” Nahmad is right that “taste is not decoration. It is direction.” Those are not indulgent slogans. They are practical strategic truths in a market where the cost of polished mediocrity is collapsing.
And finally, the industry needs to take provenance seriously. One of the clearest signals of where the market may be heading is that human authorship itself is starting to become a visible label. In March 2026, the Society of Authors launched a UK “Human Authored” scheme to help readers identify books made by humans in a market it says is increasingly flooded with AI-generated material.
The House of Lords has also called for stronger transparency obligations and for standards around provenance and labelling of AI-generated content, while the UK government says it is considering work on labels as part of its wider copyright and AI response. That does not mean every human-made thing will automatically command a premium. But it does suggest that origin, authorship and traceable human effort are becoming part of the value proposition again.
That is where this conversation gets interesting. The future probably does not belong to the people who refuse AI on principle, nor to the people who flatten their own instincts under a mountain of generated sameness. It belongs to the people who know what the tools are for, where the tools fail, and how to turn machine abundance into human distinction.
Aitkenhead says it perfectly:
“When everyone can produce content, discernment becomes the real differentiator.”
And Nahmad brings the argument to its cleanest possible finish:
“The future will not belong to those who create the most. It will belong to those who know why something should exist at all.”
That is the paradox of the age of AI. The easier it becomes to make something, the more valuable human creativity becomes when it carries judgement, originality, emotional truth and point of view. AI may keep raising the floor. But the premium will keep moving to the people who can still see the difference between output and meaning.