A mandate without a manual
Marketers are being handed a powerful tool and denied the instructions for it. New research from the market-research firm NewtonX, conducted for ADWEEK among 500 marketers, found that 69% work at companies that mandate or strongly encourage the use of AI. Yet 43% said they had received no formal training on it at all. The result is a workforce improvising with technology it was ordered to adopt, learning in public on live campaigns, and guessing at the boundary between a productivity gain and a compliance problem. A directive to use AI is easy to issue in a quarterly town hall. The harder work, teaching people to use it well, is the part most companies are skipping, and the shortfall is already shaping both the quality and the risk profile of the work marketers ship.
The proficiency illusion
The most revealing figure is what marketers think of one another. In the same study, 81% said they believe marketers overstate their AI proficiency. That is close to a professional consensus that the confidence on display is inflated, that the fluent talk about prompts and copilots runs well ahead of the skill behind it. When almost everyone in a field suspects the room is bluffing, reported adoption numbers stop meaning very much. The danger then shifts from visible failure, which a team can catch and correct, to invisible error: output that looks finished, reads fluently, and is wrong in ways no one has been trained to notice. AI is persuasive by design, which makes its mistakes harder to spot than a careless human’s, and a team that cannot spot them will keep shipping them at scale.
Why the gap is worst at the top
The shortfall does not fall evenly across the org chart. Nearly half of respondents, 47%, said senior leadership lags every other level of the company on AI. A companion analysis of the same research was blunter still, concluding that CMOs are the least proficient group in the entire marketing organization. The people signing off on AI strategy are, on this evidence, the least equipped to evaluate it, a structure almost engineered to approve the wrong tools for the wrong reasons. Outside data points the same way. Gartner has reported that only 15% of chief executives believe their marketing chiefs have the AI fluency to lead in the year ahead. When the least prepared people hold the approval authority, budgets get committed on instinct instead of understanding, and costly mistakes clear the very reviews meant to stop them.
The cost lands on the brand
The bill for untrained adoption rarely arrives as a single dramatic failure. It accumulates. It surfaces as a legal team discovering that customer data was pasted into a public model, as a campaign built on a fabricated statistic no one thought to verify, as a brand voice flattened into the same generic cadence every competitor’s tool produces. None of these register as an AI disaster in the moment. Each is a small, correctable lapse that a trained marketer catches and an untrained one waves through. Multiplied across a department using AI dozens of times a day, those small lapses become the average quality of the work. The brand ends up paying, in trust and in rework, for the training its people never received.
Training is the differentiator, not the tool
A minority of companies treat this as a solvable problem rather than an inconvenient one. Cynthia Chen, founder and chief executive of the fintech Kikoff, runs internal AI Days that teach her team not only how to use the technology for productivity but how to navigate its compliance and regulatory edges. That emphasis is the tell. Access to AI confers no advantage on its own, because every competitor has the same models a browser tab away. The advantage lies in knowing what to ask, when to distrust the answer, and how to verify it, and that judgment comes only from deliberate practice. Firms that invest in it compound small, reliable gains across thousands of daily tasks. Firms that mandate the tool and skip the teaching are buying the downside of AI, the errors and the exposure, while leaving most of the upside unclaimed.
What marketing leaders should do now
The fix is not more encouragement, of which there is already a surplus. It is structure: role-specific training tied to the actual jobs people do, a shared standard for reviewing AI output before it ships, and leaders willing to admit they need the lessons more than their juniors do. The organizations that close this gap will not be the ones that adopted AI earliest or talked about it loudest. They will be the ones that taught it best, and that treated fluency as a capability to be built rather than a box to be checked. Until then, a mandate to use AI without the training to use it well is not a strategy. It is a liability with a deadline attached.
