Cheap, Fast, and Wrong: What Advertising Loses When AI Replaces Creative Judgment

There is a strange new bargain taking shape inside advertising. Human creatives are expected to produce work that is original but familiar, surprising but safe, culturally fluent, technically perfect, immediately effective, endlessly revisable, and somehow cheaper than the last campaign. Every choice can be questioned. Every word can be softened by committee. Every imperfect frame is treated as evidence that the person behind it may not be good enough.

Then AI enters the room. It produces six fingers, drifting logos, malformed products, impossible shadows, hollow copy, and scenes whose internal logic collapses if you look for longer than a second. The response is often not rejection but accommodation: the technology is still learning. The same company that interrogates a human designer over two pixels may wave away a machine's broken anatomy as the acceptable cost of innovation.

That double standard is lowering the bar for advertising while raising the pressure on the people who know how to make it good.

AI is a production tool, not a creative department

Generative AI can be genuinely useful. It can accelerate rough concepts, widen an early exploration, organize research, test variations, remove repetitive production work, and help a small team move faster. Used by a strong creative mind, it can be another instrument in the studio.

But an instrument does not know what song should be played. AI does not understand a brand's lived history, the emotional temperature of a moment, the difference between an arresting idea and an empty stunt, or the quiet cultural signal that turns a harmless-looking image into an insult. It predicts plausible output from patterns. It does not carry taste, accountability, memory, empathy, or intent.

Those qualities belong to people. A creative director does not simply request an image. A writer does not merely arrange sentences. A photographer does not just place a subject inside a rectangle. Creative work is a chain of judgment: what to notice, what to exclude, what a choice implies, when the first idea is too obvious, when the beautiful image is wrong for the message, and when the cheapest execution will cost the brand more later.

AI can participate in that chain. It cannot own it.

The layoffs remove the very skill AI needs

The labor shift is no longer hypothetical. The Guardian reported that staffing at UK creative agencies fell by more than 14 percent in 2025, from 14,775 people to 12,659. Employment among workers age 25 and younger fell 19.2 percent. Advertised jobs dropped 41 percent across the industry, with the decline at creative agencies approaching half.

The most telling number may be the future-facing one: only 43 percent of agencies reported taking on trainees, apprentices, or school-leaver apprentices, down from 56 percent a year earlier. That is not only a headcount reduction. It is a break in the system that creates experienced creative judgment in the first place.

No senior art director begins as a senior art director. Taste is developed through briefs, failures, edits, production constraints, difficult feedback, client conversations, and years of seeing why one almost-correct solution still does not work. When businesses automate the junior work and stop hiring the junior people, they do not merely save today's salary. They erase part of tomorrow's leadership pipeline.

A Stanford Digital Economy Lab study found employment declines concentrated among early-career workers in occupations highly exposed to generative AI, while more experienced workers were comparatively protected. The researchers explicitly point toward the value of tacit knowledge: the uncodified, experience-built judgment that AI is less capable of replacing.

In other words, companies are cutting the people who need opportunities to develop the exact quality the machine lacks.

The machine gets grace. The human gets a performance review.

When a human-made campaign misses, the critique is often forensic. Why was this approved? Who wrote this? Did nobody catch the continuity error? Was there no one in the room who understood the audience?

Those are fair questions. Advertising reaches millions of people and spends real money; standards should be high. The problem is not that human work receives scrutiny. The problem is that AI work is too often granted a lower standard precisely when it is being sold as the replacement.

Coca-Cola's 2024 AI-generated holiday campaign drew widespread criticism for visuals viewers described as soulless and for the uncanny inconsistencies associated with generated imagery. NBC News reported that the spots were produced by three AI studios using four generative models. That is not the story of a machine independently making an advertisement. It is the story of a production system with many points at which a human could have asked whether speed and novelty were improving the work—or merely making its weaknesses more expensive and more visible.

A brand would not accept a photographer saying that distorted hands are simply where the camera is right now. It would not excuse a copywriter because coherent language is difficult. It would not tell customers to admire the potential of a broken product photograph. Yet attach the word AI and obvious flaws are recast as evidence of experimentation.

Innovation cannot mean charging the audience for a beta test.

The cement-mixer treadmill

The recent Nike Sportswear campaign Above the Influence offers a useful complication. One image shows a man walking on a cement mixer as if the rotating drum were a treadmill. Other scenes place people in similarly impossible situations. At first glance, it looks like the visual grammar we have learned to associate with AI: physically plausible enough to stop the scroll, logically unstable enough to make us ask what we just saw.

But reporting on the campaign identifies it as an intentionally surreal concept directed by human creative Pablo Rochat for Nike's Air Max line. Collater.al describes the campaign as a series of visual short circuits built from absurdity, irony, and dry humor. The cement mixer is not an AI mistake. The lack of realism is the idea.

Whether that idea works is a different question. Absurdity can be sharp, funny, and memorable. It can also become a closed loop in which attention is the only meaning: the image stops us because it is strange, and it is strange because stopping us was the entire brief. The brand gets a moment of confusion, the platform gets another replay, and the audience is left trying to locate the point.

That is why creative judgment matters even when no AI is involved. A technically intentional image is not automatically a meaningful one. A creative mind must still ask what the absurdity reveals about the product, the audience, or the culture. If the answer is only people will look twice, advertising has mistaken interruption for communication.

The campaign also reveals how quickly AI has changed visual literacy. Audiences now encounter so many synthetic impossibilities that an intentionally constructed surreal scene can be read as another machine hallucination. When every feed is full of frictionless strangeness, strangeness itself loses value. The challenge for creatives becomes harder: not simply to make what has never been seen, but to make something whose reason for existing can still be felt.

Businesses are not saving what they think they are saving

Replacing a creative team with generation tools can reduce an invoice. It can also create costs that spreadsheets hide.

• Sameness. Models are built from patterns, and unchallenged output drifts toward the statistical center. Brands that automate the thinking begin to sound and look like every other brand using the same tools.
• Rework. Fast generation is not fast approval. Someone still has to inspect claims, continuity, anatomy, typography, product accuracy, legal risk, accessibility, and cultural context. Remove the expert reviewer and the errors travel downstream.
• Trust. Audiences recognize when a company invested less care in speaking to them. The damage is not always a viral backlash. More often it is a quiet reduction in credibility, attention, and emotional connection.
• Institutional memory. Freelance prompts and generated variations do not replace the person who remembers why a campaign failed three years ago, how the audience talks when no one is selling to them, or which visual shortcut contradicts the brand's values.
• Capability. A company that removes its creative bench becomes dependent on tools it cannot effectively direct or evaluate. It may generate more assets while losing the ability to tell which ones are good.

The output count rises. The standard falls. The dashboard can make that look like productivity right up until the brand becomes forgettable.

The better model is creative coexistence

The answer is not to ban AI from the studio. Refusing a useful tool is no more strategic than replacing the studio with it.

The better model keeps human creatives in authority. Let AI expand possibilities, speed repetitive work, generate disposable drafts, and help teams reach the first conversation faster. Then let experienced people interrogate every result. Does it make sense? Is it true? Is it ours? Is it respectful? Is it useful? Is it interesting for a reason beyond the fact that a machine made it?

This is not ceremonial human oversight—the final approval click after the direction has already been automated. It is human authorship throughout the process: defining the problem, shaping the concept, choosing the references, challenging the first output, refining the language, art-directing the details, and taking responsibility for what reaches the public.

Creative people are not valuable because they can move a cursor faster than a model can render. They are valuable because they can recognize the distance between an image and an idea.

Raise the standard for both

Human-made work should be questioned. AI-assisted work should be questioned just as hard. The standard cannot depend on which tool produced the first draft; it has to depend on what the final work asks the audience to accept.

If a campaign is incoherent, the technology is not an alibi. If an image is wrong, cheapness does not make it right. If a brand removes the people capable of critique, it has not built an AI strategy. It has dismantled its quality control.

The future of advertising does not belong to humans pretending AI has no value, or to businesses pretending AI has taste. It belongs to creative people who know how to use powerful tools without surrendering judgment to them.

AI can make more. A creative mind still has to decide what is worth making.

Sources

Collater.al. (2026). Nike's strange (and funny) campaign by Pablo Rochat. https://www.collater.al/en/nike-above-the-influence-pablo-rochat

The Guardian. (2026). UK ad agencies undergo their biggest exodus of staff as AI threatens industry. https://www.theguardian.com/media/2026/feb/13/uk-ad-agencies-biggest-annual-exodus-of-staff-ai-threatens-industry

Brynjolfsson, E., Chandar, B., & Chen, R. (2025). Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence. Stanford Digital Economy Lab. https://digitaleconomy.stanford.edu/wp-content/uploads/2025/08/Canaries_BrynjolfssonChandarChen.pdf

NBC News. (2024). Coca-Cola causes controversy with AI-made ad. https://www.nbcnews.com/tech/innovation/coca-cola-causes-controversy-ai-made-ad-rcna180665