Mental Health Awareness Month is here, and the topic leading the agenda isn’t about removing the stigma around mental health (we’re arguably moving past that), it’s whether AI is making mental health worse.
Research from Upwork found that while AI users reported a 40% productivity boost, 88% of the most productive AI-enabled workers also reported burnout, and they were twice as likely to be considering leaving their jobs. A BCG study published recently linked what researchers are now calling “AI brain fry” to increased fatigue and significantly higher quit intent: Thirty-four percent of workers experiencing it said they planned to leave, compared to 25% without it.
So, while we may be getting more done, the people doing it are paying a price that doesn’t show up in any dashboard.
The fault lines are pretty clear when you take a closer look at the process.
When employees use AI for their work, they don’t just complete tasks. They’re in a hypervigilant state as they continuously monitor the output: Is it accurate? Is it compliant? Does it conflict with what another system told me 10 minutes ago? Can I actually act on this, or do I need to verify it first?
Our research at WalkMe calls this decision latency. It happens dozens of times a day, in the pause before each AI output is accepted or rejected.
And it’s not irrational caution. It’s the reasonable response of employees who have been given powerful tools without the guardrails or guidance to leverage them.
In our 2026 State of Digital Adoption report, only 12% of workers said they are fully confident that AI tools understand the context of their work. That’s the root of that relentless hypervigilance. The technology is working. But the trust just isn’t there.
See also: ‘Digital fluency’: What it is and why business leaders value it so much
The dangerous assumption of fluency
In the void of proper training, many employers are expecting their employees to just figure it out. For some, it’s out of avoidance or not knowing what resources to turn to. For others, it’s because they have the resources and permission to learn it themselves, driving a dangerous assumption of fluency. It’s important to remember, though, that just because AI is intuitive for some, that doesn’t mean it’s intuitive for all. Nonetheless, this faulty belief is shaping rollout decisions in ways that leave the workforce underprepared and unprotected.
Our data reflects this: Eighty-eight percent of leaders in the WalkMe study believe employees have adequate tools. Only 21% of employees agree. They are describing different workplaces.
And when the approved tools create too much friction, employees find alternatives. Our research revealed that 45% of workers used unapproved AI tools in the past 30 days, and 36% of them did so with confidential data.
This is routinely framed as a compliance failure when, in fact, it’s a design failure.
What leaders are still getting wrong
Organizations have been measuring AI success by adoption rate and deployment scale. However, the question that matters for employee experience and sustainable performance is not “How many employees are using AI?” but “How well are we setting them up to use it?” Cognitive load, decision fatigue, rework and the anxiety of operating in a policy gray zone are all real operational costs. They just don’t appear on the same reports as efficiency gains.
The shift that’s required is from transformation (setting the strategy) to adoption (delivering the results). One is announced. The other is built, system by system, workflow by workflow, in the moment when an employee needs guidance and gets the tools they need to learn and develop.
Practically, that means HR needs to lead on a few different fronts and across functions to effect change. First, partner with IT to measure where work is actually breaking down, not just whether tools are deployed. Second, rethink the training model entirely. Traditional learning and development can’t keep pace with the way organizations are deploying and updating AI. Our research found that workers who receive training and support within the tools and workflows they are already using are 3.7 times more likely to gain confidence in using the tool effectively. And lastly, make this perception gap visible to leadership as a retention risk. This is HR’s business case to own. That 88% to 21% gap in confidence means you are already burning out your top performers.
This Mental Health Awareness Month, along with the conversations about benefits and support resources, it’s time to address this issue as well. The way AI is currently being deployed is increasing the mental load on the very people organizations are counting on. Finding the solution is both a human imperative and a business one.
Credit: Source link
