The lying has slowed down. The overconfidence has not.
That is the blunt takeaway from WalkMe's third annual AI at Work Pulse Survey, released Tuesday and conducted by Propeller Insights among 2,037 working U.S. adults. The share of employees who admit to pretending to know AI in a meeting fell from 45.2% in 2025 to 28.3% in 2026. The share who admit to passing off AI-generated work as their own dropped from 48.7% to 32.5%.
Those numbers look like progress. WalkMe cofounder and CEO Dan Adika reads them differently.
'Workers are not becoming more deceptive; they are becoming more comfortable,' the survey release states. Adika's translation: nobody needs to fake competence they sincerely believe they already have. Ninety percent of workers say they feel confident using AI. Only 24.6% say it works on the first try. Half say they have spent more time trying to get AI to complete a task than the task would have taken manually.
Adika offered a sharper test at a major SAP conference in Madrid in May. He asked a room of enterprise software executives to raise their hands if they were using AI in a meaningful way. A recent Gartner survey put that figure at roughly 8%. 'It was only two,' he told Fortune. 'That's the reality.'
The gap between sentiment and results becomes a financial question the moment a CFO opens a spreadsheet. 'If you have 50,000 employees, so you should save 50,000 hours, let's call it a week. So you should have saved 200,000 hours a month. Where are the $4 [million] or $5 million in savings?' Adika asked. His answer: 'It is not there. It's not translating to actual P&L savings.'
The confidence problem runs to the top of the org chart. The survey finds 53.6% of employees say they have felt a senior leader does not fully understand the AI strategy that leader is publicly championing. 'Everyone, from the newest hire to the executive suite, is learning AI in real time,' said WalkMe Global Field CTO KJ Kusch.
Adika also flagged a double standard that slows adoption further. 'Everybody expects AI to be perfect,' he said. 'If AI is wrong 1%, while a human being is wrong 10%, all the focus would be, wow, the AI got it wrong — but he got it wrong 1% while everybody else gets it 10%. There is a big resistance.'
The sharpest warning in Adika's account concerns what companies are actually asking workers to do right now: encode their own expertise into AI systems he calls a 'company brain.' The math that follows is uncomfortable. A manager who once needed a team of 10 to execute decisions can, once the system is trained on that team's collective knowledge, operate with two decision-makers and one person checking the AI's output. 'So now you can shrink the team from 10 to three,' Adika said.
That leaves the employees doing the training in a bind. 'They take all their knowledge, they move it to the AI. Now the AI can do it instead of them. Now they might fire them, right? So it's a catch,' he said. 'No one has a good answer for that.'
CEO Times take: The survey data confirm what free-market observers have argued from the start: adoption metrics divorced from productivity metrics are marketing, not management. Capital does not reward confidence surveys; it rewards margin expansion. Until AI investment shows up on the P&L — not in employee sentiment scores — enterprises are paying for a feeling. The workers being asked to train their own replacements deserve straight answers from leadership, not another round of mandatory training modules. Clear rules, honest accounting and respect for the individual's stake in their own expertise are the baseline. Right now, too few boardrooms are meeting it.
