Enterprise & Work

68% of managers require AI use, but employees differ on dependence

A Slingshot poll of 500 U.S. employees and managers finds AI mandates rising even as workers report more choice and less dependence than leaders expect.

More Managers Require AI at Work, but Employees See Its Role Differently
More Managers Require AI at Work, but Employees See Its Role DifferentlyStartupStockPhotos / pixabay
By Elena Vasquez6 min read

Why it matters

  • 68% of surveyed U.S. managers required AI use, up from 59% in the previous edition.
  • 51% of managers expected slower work without AI; 26% of employees said the same about their own work.
  • Two-thirds of employees used AI without a requirement; another 23% would use it anyway.
  • Research and summarization led employee uses at 57%, ahead of drafting at 45%.
  • 42% of managers said their organizations adopted AI for decision-making, up from 20% in 2025.

By Elena Vasquez

Sixty-eight percent of U.S. managers in Slingshot’s 2026 Digital Work Trends Report said they require employees to use AI, up from 59% in the previous edition. The nine-point rise shows tighter mandates. Slingshot’s report finds a different employee view of dependence.

About half of managers, 51%, expected employees to struggle or work more slowly without AI. By comparison, 26% of employees voiced that concern about their own work.

That gap leaves a practical management problem: leaders can mandate a tool without knowing which tasks workers could still complete at the same pace without it.

The two percentages answer different questions. They do not directly compare managers with their own reports. Still, Slingshot presents the contrast as evidence that visibility inside a workflow does not equal dependence.

What does the mandate gap mean?

Slingshot is a work management platform from Infragistics. It published the first installment of its 2026 Digital Work Trends Report on November 19. Dynata, the market research firm, collected the responses in the fourth quarter of 2026.

The sample contains 500 U.S. participants. That includes 300 employees and 200 managers. The fieldwork covered four age groups and all 50 states. Slingshot commissioned the research.

The report measures what participants said about AI. It does not show how much work they completed or whether their decisions improved. That limit matters because the survey cannot establish productivity gains.

Managers’ expectations therefore carry more weight than a simple adoption count. Slingshot says leaders can see how often a tool appears in a workflow without knowing what workers could do without it. The survey suggests mandates and practical dependence have started to pull apart.

That gap remains useful, although the two groups answered different questions about hypothetical productivity rather than reporting directly on the same manager–report pairs over time.

Do employees choose AI without pressure?

In Slingshot’s survey, two-thirds of employees said they use AI without a company requirement. Another 23% said a requirement applies, but they would choose to use the tool anyway.

Those responses put voluntary use at the center of the employee picture. Slingshot’s data do not support a simple story of reluctant workers pushed into software. They support a more complicated one: employers mandate AI while many workers have already made it part of their own workflow.

Generational expectations split again within that picture. Forty-six percent of Gen Z employees said they would struggle or slow down without AI. That is higher than the 26% reported by employees overall.

The published summary does not provide a headcount for the Gen Z group. Slingshot therefore presents the 46% as a signal to investigate, not a precise estimate for younger U.S. workers nationwide.

Still, the finding shows why one company mandate cannot stand in for one workplace dependency profile. Slingshot treats use as personal, task-specific and already voluntary in many cases.

What do employees use AI for?

Slingshot found that research and summarization ranked first among employee uses. It reached 57% in the 2026 report, up from 48% in the prior edition. The gain makes information work the clearest category for employers to examine.

Slingshot also found that 45% of employees cited drafting written content. Another 41% identified drafting emails, reports or other written material as the work they most often keep manual. The numbers show both adoption and a deliberate boundary around authorship.

The results are not mutually exclusive. One employee can ask AI to pull together background for a report, then write the report without it. Slingshot’s framing allows research assistance and manual drafting to sit in the same workflow.

That distinction undercuts any claim that an employee either works entirely with AI or entirely without it. The survey records common tasks instead. It does not claim that AI creates the final research, report or email.

Those task choices fit the report’s explanation for why managers may overestimate dependence. A tool can be useful at the start of a process and still leave the final production step firmly in human hands.

Does AI actually save time?

A separate Gallup survey of 21,648 U.S. employees provides a useful comparison. Among frequent AI users, 52% said they spent less time gathering information after adopting AI.

Across Gallup’s broader sample, 32% used AI at least a few times a week. Gallup’s figure cannot be read directly against Slingshot’s employee-use measures. The surveys asked different questions of different groups. Their numbers are not directly comparable.

Read together, the Gallup and Slingshot results point to a productive place to test AI’s labor effects, while the different samples also show how adoption figures change with the population, question and definition of use.

The Gallup result links AI with less time spent gathering information among frequent users. The Slingshot result places research and summarization first among employee use cases. Together, they support studying time saved at the information stage. They do not establish that every organization will see the same reduction.

Frequency also matters in the comparison. Gallup measured a subset of frequent users for the time-saving result, then reported use at least a few times a week across its broader sample. Slingshot measured reported workplace roles and expectations. Neither survey tracked the same employees before and after deployment.

What should employers measure next?

Slingshot reports that 42% of surveyed managers said their organizations brought in AI tools to improve decision-making. That figure rose from 20% in 2025.

Slingshot’s 42% figure records the reason for adoption, not the quality of the result. The report draws a sharp line between bringing in a tool and making a better decision. The second still requires evidence.

The report describes a clear trade-off. An AI summary can save time early in a task, then create more work later if reviewers must chase missing references or correct confident errors. Speed at first response can conceal a slower verified process.

For a recurring research task, Slingshot recommends comparing “the time from assignment to verified output, including review and rework.” Its broader guidance is equally direct: “Employers should measure verified output time instead of counting prompts or adoption rates.”

Employers should not count prompts or measure how quickly the first answer appears. The endpoint is a checked result, not an instant suggestion. That approach includes the review burden that an adoption rate leaves out.

For a decision based on an AI-generated summary, Slingshot says to check whether a colleague can trace the recommendation to the underlying evidence. That test exposes missing support before a manager acts. It also makes review part of the measured workflow, rather than an invisible cleanup step.

Those measures will become more useful as workplace AI becomes routine. A mandate tells leaders that a policy exists. An adoption rate tells them that people used a tool. Verified cycle time, review burden and traceability tell them whether the work improved.

Slingshot’s advice gives employers a starker test than counting AI users. Measure how long a correct, reviewable result takes. Then check whether anyone can show the evidence behind it. As workplace use becomes routine, those tests could decide whether AI mandates improve productivity or merely standardize its presence.

Source: unite.ai

Share this article:

More from Elena Vasquez

Elena Vasquez

Show full bio

Market editor covering technology and product news in artificial intelligence and machine learning at AI In Context.

795 articles

Related articles

  1. AI Agents Become Coworkers—and Managers of the Bots
  2. Australia Leads AI Agent Adoption at 46%, but Managers Pay the Price
  3. AI Literacy Is Not a Training Course: Why Governance Decides Who's Ready
  4. 67.8% of Workers Now Use AI Weekly — But 56.4% Get No Time to Learn It
  5. APA Tells Employers: AI Deployment Is a Job Design Problem, Not a Purchase

« Previous articleNext article »