Adland has decided taste and judgement are its last human advantage. Nobody is asking where the next generation is supposed to acquire these skills now that AI has absorbed the junior roles that used to teach them, writes Cat McGinn, founder of humAIn.
Many of the AI conversations this year have centred on judgement, taste, and craft as the uniquely human advantage; a reassuring lullaby for an industry trying not to have nightmares. But this becomes a platitude if we don’t also reckon with how judgement is formed, and craft is honed.
That is the uncomfortable truth arising from D&AD’s second annual AI & Creativity Report, The Cost of the Shortcut. Last year, the industry was still debating whether AI could ever become truly creative. This year, after 197 conversations with creative leaders across 30 countries and analysis of 10,118 D&AD Awards entries, that question has disappeared.
What has emerged instead is something nobody was worrying about last year: the atrophy of judgement. D&AD’s 2025 report ended with the proposition that AI didn’t unlock creativity; it exposed who lacked it.
This year the report opens with a firmer conclusion: “AI didn’t unlock creativity. It exposed the value of judgement.” And this year, there is more data.
AI-declaring entries to the D&AD Awards more than doubled year-on-year, from 10.7 per cent to 27.6 per cent. AI work won trophies at roughly the same rate as the overall pool, but generative AI accounted for 56.2 per cent of AI-declaring entries and only 44.7 per cent of AI Pencil winners. The stronger work, it seems, was more likely to have used AI to enhance the idea than to have used it as the source.
The problem is not that AI cannot produce competent creative work. It is that it can.
As the report’s author Ben Cooper says, “AI is brilliant at removing friction from creative work, and friction is where judgement is built. Take out the slow, unglamorous parts and you also take out the place where people learned the difference between an answer and an idea.”
Or, more precisely, between a plausible answer and a distinctive idea.
Judgement has a supply chain. It starts with research, bad first drafts, layouts, rejected ideas, revisions, critique and proximity to people who know more than you do. We used to excuse the paucity of structured learning and development by saying that you’d learn by osmosis; just being around your creative elders would somehow allow their expertise to seep into your marrow. And actually, that proved to be largely the case, although it was never evenly distributed.
Now, our disconnected and disaffected workforce, compounded by the rise of AI, is losing the repetitive work that was also a de facto apprenticeship.
The supply chain itself is being cut.
D&AD cites IPA data showing the proportion of employees aged 25 and under at its member agencies has fallen sharply, while entry-level hiring has slowed. AI isn’t being used as a fig-leaf: only 8 per cent of UK agencies directly attribute recent headcount reductions to it. Consolidation, margin pressure and wider structural changes are the bigger influence.
Whatever the cause of fewer junior jobs, AI is absorbing many of the tasks through which juniors traditionally learned, and at the same time, there are growing signs of convergence in the work itself.
Young Lions judge Tim Collier reported that repetitive outcomes made AI-assisted submissions recognisable, while Ant Melder described entries where AI had generated the idea as among the most “generic and expected”.
That cuts both ways: machines are tending towards plausible sameness, while people become more accepting of average outputs.
Wharton researchers Steven Shaw and Gideon Nave describe this as “cognitive surrender”: we accept an AI-generated answer with minimal scrutiny, rather than making the effort to use AI while retaining judgement.
That is why D&AD’s warning about junior work matters beyond the employment question. If judgement is built through trying, failing, rejecting and revising, then removing those opportunities changes the capability we are producing. We are automating part of the apprenticeship while continuing to assume mastery will somehow keep arriving at the other end.
And audiences can tell. When people can see that content is AI-generated, 45 per cent of Australians reporting they trust a brand less.
Which makes this, ultimately, a leadership problem. The D&AD report describes leaders rolling out tools they do not fully understand to teams with little guidance about how they should be used. One creative director says leadership “can only guess” at what AI means for headcount, while morale suffers because teams can see that uncertainty.
Organisations are asking people to experiment with unfamiliar technology while many of those same people are anxious about what that technology means for their jobs. A climate of fear does not create the conditions for useful experimentation, much less creativity.
People who believe a mistake may become evidence of their expendability do not readily show the bad first draft, try the strange route, admit uncertainty or challenge the machine’s plausible answer. They optimise for looking competent.
And competence is precisely what AI can already produce, cheaply.
Scared people make safe work. If judgement really is the scarce human capability of the AI era, building psychological safety is not merely a soft skill to be delegated to the HR department. It is part of the infrastructure through which judgement develops.
The leadership task is not to preserve every inefficient way of working. It is to understand which friction was merely waste and which friction was formative, and to deliberately replace the latter rather than automating it away. To audit tasks and determine what is scaffolding, and what should not be automated. Pairing juniors with senior talent. Measuring not only efficiency gains, but capability built.
In an industry already prone to navel-gazing, we are becoming brilliant at identifying the capabilities we want to preserve, while paying much less attention to the conditions required to reproduce them.
D&AD’s 2026 report turns that tension from a comforting mantra into a harder organisational question. If judgement is the thing the industry plans to charge a premium for, who is taking responsibility for replenishing the stock?
