A new APA survey shows AI use rising alongside worker concern. Resilience matters, but leaders still owe people truthful communication, real preparation, and visible ownership.

The American Psychological Association released its 2026 Work in America survey today, and the AI findings reveal a problem many leadership teams are still treating as an attitude issue.

Nearly half of the workers surveyed—48%—said they worry AI could make some or all of their duties obsolete. Sixty-four percent said technological change primarily benefits employers rather than workers, up from 52% in 2025. Another 37% worried they would fall behind colleagues if they did not use AI, compared with 30% last year.

At the same time, APA reports that intentional daily use of AI at work has continued to rise.

That combination matters. People are not necessarily rejecting the technology. Many are using it while remaining uncertain about what its adoption means for their jobs, their opportunities, and their standing inside the organization.

The leadership failure would be to interpret that uncertainty as a motivation problem.

Resilience cannot carry an unexplained system

APA’s central frame is “psychological capital,” a research construct combining hope, efficacy, resilience, and optimism. In the survey, workers with higher scores were less likely to see AI as a threat and more likely to report that technological change had created opportunity in their work or career.

That is worth paying attention to. People need the capacity to learn, recover, solve problems, and move through uncertainty. No organization can remove every fear before work changes, and no employee is relieved of the responsibility to grow because a transition is difficult.

But the survey does not prove that psychological capital causes better outcomes. It reports associations from an online poll, and the conditions around a worker matter too. Hope is not a substitute for information. Resilience is not a substitute for preparation. Optimism cannot answer a question leadership refuses to address.

If an employee believes the institution will gain while the worker carries the risk, another workshop on embracing change will not repair the trust gap. Leaders have to explain the transformation itself.

Tell people what is actually changing

Many organizations communicate AI adoption in one of two bad ways. They announce a grand transformation before anyone can describe the work, or they quietly introduce tools and leave employees to infer the strategy from scattered pilots and rumors.

Neither approach gives people a truthful account of what is happening.

A responsible explanation does not require a five-year forecast. It requires leaders to distinguish what they know from what they do not know. Which workflows are being tested? What problem is the technology supposed to solve? Which decisions remain human? What information may employees put into the system? How will output be reviewed? Which roles are likely to change first? What has not been decided?

That last question may be the most important. False certainty destroys credibility when reality changes. Honest uncertainty can build trust when it comes with a clear process for learning and deciding.

Employees do not need leaders to predict every consequence. They need leaders who will not hide behind the uncertainty while continuing to make consequential choices.

Preparation has to be specific to the work

“Learn AI” is not a training plan.

People need a shared language for what responsible use means inside their organization: which sources are authoritative, what quality looks like, when disclosure is required, what information is prohibited, which errors matter most, and who can stop or escalate a questionable result.

Then training has to reach the actual role. A manager deciding how to allocate work needs something different from a writer preparing a public claim or an analyst validating a recommendation. Tool familiarity helps, but adoption depends on whether people can apply judgment in context.

Workers still have responsibility here. They should learn the tools that are becoming relevant to their work, ask better questions, test their assumptions, and refuse the comfort of saying, “That is how we have always done it.” Protecting an old method is not the same as protecting the mission.

Yet leaders cannot demand adaptation while leaving people to guess at the rules. If using AI is becoming part of the job, time to learn, access to approved tools, examples of acceptable work, and permission to surface problems have to become part of the system.

Ownership must remain visible

AI programs often become strangely ownerless. Executives approve the direction. Technology teams configure the tools. Managers push adoption. Employees carry the daily decisions. When something goes wrong, responsibility moves toward the person closest to the output.

That is backward.

A leader still owns the organizational decision even when a worker uses the tool. Ownership should be visible in the policy, the training, the workflow, the escalation path, and the way gains and burdens are measured. If a new system saves management time but adds verification, correction, or anxiety elsewhere, that work belongs in the evaluation.

The aim is not to promise that every role will remain unchanged. No honest leader can make that guarantee. The aim is to deal truthfully with people whose work is being redesigned and to prepare them for choices the organization is actively making.

The APA survey offers a snapshot, not a diagnosis of every workplace. The Harris Poll surveyed 2,009 employed U.S. adults online from May 21 through June 1, 2026. The results were weighted to population characteristics, with a reported Bayesian credible interval of plus or minus 3.1 percentage points for the full sample; subgroup intervals are wider, and the usual limits of online, self-reported survey research apply.

Still, the pattern deserves a direct response. AI use and worker concern can rise together. When leaders do not explain why the technology is being introduced, people may learn the tool while doubting the institution’s account of who will benefit.

Leaders should expect workers to learn and adapt. They should expect more from themselves first: explain the real change, prepare people for the work, establish a common language, and own the consequences in public.

You cannot coach people into trusting a transformation you have not explained.

Sources