TeamViewer: An "AI Confidence Gap" Is Quietly Slowing Workplace Progress
TeamViewer has released new global research arguing that the barrier to the next wave of AI productivity is no longer adoption — it is trust. The study, Path to the Autonomous Digital Workplace, finds that 75% of respondents already use AI at least once a day, yet 61% still prefer that AI take no independent action on their behalf.
That reluctance sits awkwardly against enterprise ambition. IT leaders surveyed expect close to 40% of digital workplace services to be running autonomously by 2030 — a trajectory that assumes a level of delegation most employees have not yet signed up for.
"Trusted autonomy, not unchecked automation."
— Oliver Steil, CEO, TeamViewerComfort With Assistance, Caution With Authority
The pattern TeamViewer labels the AI Confidence Gap is straightforward: people are relaxed about AI that helps, and far more guarded about AI that decides. Sentiment toward the technology itself is broadly warm — 64% describe their experience with AI as positive, and only 3% report no noticeable workplace benefit at all. Confidence erodes as the system gains access, authority and responsibility.
Even among the 35% who say they are ready to let AI act more independently, most are hedged: 71% of that group are only comfortable with narrowly defined tasks. Just 5% report no concerns whatsoever about AI operating without a human in the loop.
The Verification Tax
The gap carries a measurable productivity cost. More than half of respondents — 56% — often or always check AI outputs before relying on them, spending an average of two hours a week reviewing AI-generated work. Compounding it, 51% say they do not reliably know when to trust an output and when to verify it. The time AI saves at the point of creation is being partly handed back at the point of review.
The design challenge ahead is software that recognises when to act on its own, when to hold back, and when a person needs to be brought into the decision.
— Mei Dent, Chief Product & Technology Officer, TeamViewer (paraphrased)Visible Autonomy, Not Invisible Autonomy
Crucially, the research does not describe a workforce opposed to automation. It describes one opposed to automation it cannot see. 70% are comfortable with AI acting on its own provided they can intervene when needed, and 40% actively want AI to catch and fix problems early while keeping them informed.
Respondents were specific about what would make autonomous action acceptable: strong security and privacy protections (39%), notification before major changes (36%), clear limits on what AI can access (36%), visible activity logs (33%) and the ability to roll changes back (31%). Guardrails, in short, are not friction — they are the precondition for scale.
IT Is Where Autonomy Lands First
IT operations look set to be the proving ground. 26% of IT leaders already describe their departments as at least partially autonomous, and virtually all of them — 99% — would hand at least one task to AI today. The most commonly cited candidates are system updates (61%), routine device fixes (53%) and device performance tuning (49%): work that is repetitive, measurable and easy to reverse.
As for the anxiety that usually shadows this topic, the study is comparatively calm. Looking ahead to 2030, 33% expect AI to assist while they continue doing most of their tasks, 28% expect to delegate more and move up to higher-value work, and only 2% expect their role to disappear entirely.
The research was conducted by Sapio Research on TeamViewer's behalf in April and May 2026, covering 4,200 respondents across nine markets — the U.S., U.K., Australia, Brazil, France, Germany, Italy, Japan and Singapore — with an even split of managers and employees across industries and company sizes.
Adoption is solved; trust is not. 75% use AI daily and 64% rate the experience positively, yet 61% still want it to take no independent action without a human involved.
Ambition is outrunning confidence. IT leaders expect nearly 40% of digital workplace services to run autonomously by 2030 — well ahead of where employee comfort currently sits.
Verification is eating the gains. 56% routinely check AI outputs, averaging two hours a week, and 51% are unsure when trust is warranted and when verification is needed.
The objection is to invisible autonomy. 70% accept AI acting on its own if they can step in, and guardrails — security, notifications, access limits, logs, rollback — are what make that acceptable.
IT is the first proving ground. 99% of IT decision-makers would delegate at least one task now — updates, device fixes, performance tuning — while only 2% of workers overall expect their role to be eliminated by 2030.