The 1,002 US workers in Adobe’s survey all use AI weekly, yet only 9% use it mostly for creative work, and 47% say the reason is that nobody has shown them how it applies to their role.

Adobe’s newest workplace survey has produced one number that travels well and one finding that matters more. The number is 400 hours. The finding is that nearly half the workers who already use AI every week say nobody has shown them how creative AI applies to their job.

The distinction matters because marketing teams are being asked to justify AI spending in budget rounds where every tool needs a business case, and a figure describing hundreds of hours of waste is the kind of thing that ends up on a slide. Adobe’s study does not support that reading. What it does support is cheaper and more useful: the main obstacle to wider creative AI use is that people cannot see where it fits their work. That is a briefing problem rather than a procurement one, and it sits underneath the same adoption-versus-outcome split The Helm has tracked through 2026.

What did Adobe actually measure?

Adobe surveyed 1,002 full-time US workers who already use AI at work at least once a week, then asked them about creative uses specifically. The study, published on Adobe’s Acrobat resources site as “The AI Use Iceberg”, separates creative uses — image and video generation, designing graphics, brainstorming, creative writing, visualising concepts — from technical ones such as summarising documents, drafting content, research, data analysis and workflow automation. Ninety per cent of that sample said they were interested in using creative AI at work. Nine per cent said they use AI mostly for creative purposes.

Those two figures are not opposites. The 9% measures which mode of use dominates, not whether creative AI gets used at all. Adobe’s own data shows 69% of the same group already use AI for brainstorming, an activity the study defines as creative in one section and lists among “practical tasks” in another.

The 400 hours has a similar shape. Respondents estimated they spend 7.5 hours a week on tasks that could be enhanced by greater use of creative AI, and Adobe multiplies that across a year to reach “nearly 400 hours”. It is a self-reported estimate of addressable time, not measured waste and not measured saving. Adobe’s methodology note states the research is exploratory, relies on self-reported data, and that discrepancies may exist between what respondents said and what they actually do. The study carries a 95% confidence level and a margin of error of roughly 3%. Trade coverage has since framed the figure as hours workers waste, a word the study does not use.

Why do workers who want creative AI not use it?

Adobe’s barrier ranking puts relevance well clear of everything else. Forty-seven per cent of respondents said they are unsure how creative AI applies to their role. Lack of training or resources and unclear company policy tied at 29%, difficulty integrating with existing workflows reached 27%, cost 26%, and lack of access 24%. Time constraints came in at 20%. Resistance from colleagues or management finished last, at 13%.

Read in order, that list argues against the usual explanations: cost and access sit in the lower half, and reluctance sits at the bottom. These workers are not holding out; they use AI weekly by definition of the sample, and one in three said they would like to generate images or video at work but currently do not.

What respondents said would shift that is equally plain. Examples of how AI applies to their specific role topped the list at 37%, followed by better integration with existing tools and workflows (36%), recommendations from colleagues or industry experts (34%), employer-provided training or onboarding (33%) and data privacy and security assurances (32%). One in three had received formal training on creative AI from an employer, while 53% said their employer supports learning it.

None of the top three items requires a purchase order. A single worked example — a campaign concept taken from brief to visual, shown by a colleague rather than a vendor — answers the first and third at once, and costs an afternoon.

Does independent research point the same way?

SHRM’s Navigating AI in the Workplace study, fielded to 5,875 US workers in March and April 2026, approaches the enablement question from a far broader base and lands in a similar place. Just under 40% of workers said their employer had run workshops on practical, day-to-day AI skills, and 37% had received an introduction to AI tools. Among organisations that had adopted AI, fewer than half (47%) had a policy governing its use, and 46% of workers said existing policies stopped them experimenting with new tools. McKinsey research presented at Cannes Lions in June found the same shape at organisational level: enthusiasm near-universal, value capture in single digits.

The two datasets differ on scale in a way worth holding on to. SHRM found 41% of its sample use AI for work at all. Everyone in Adobe’s sample uses AI at work weekly, which makes them an unusually engaged group, and the enablement gap shows up even there.

Adobe also has a commercial interest in the answer. The study sits on a page that links to Firefly and Acrobat, and it follows a run of similar Adobe workplace surveys published through 2026. That does not make the barrier ranking wrong; vendor-run surveys can produce useful data, and this one is more candid about its limits than most. It does mean the conclusion Adobe draws from it, that its own products bridge the gap, is marketing, and the barrier ranking is the part worth keeping.

Adobe’s study page carries no publication date and no link to the underlying tables. The survey covered US workers only.