Your team is using AI. Employees are enthusiastic. Hours are being saved. So why isn’t productivity actually improving?
New research from Workday has uncovered what the company is calling “the AI tax”, and it’s costing businesses more than they realize. For every 10 hours employees save using AI tools, nearly four hours vanish into fixing errors and fact-checking outputs that can’t be trusted.
The study of 3,200 employees across global enterprises found that 85% report saving one to seven hours per week with AI, yet only 14% consistently achieve positive net outcomes. The difference is what Workday terms “rework”.
The ‘Low-Return Optimist’ Problem
The people most hurt by AI rework are seemingly the ones who believe in it most.
Daily users of AI are overwhelmingly optimistic, with more than 90% believing these tools will help them succeed. Yet Workday found that 77% feel compelled to review AI-generated work more carefully than they would human output.
The AI productivity paradox isn’t a problem with the technology itself. While 66% of leaders say AI skills development is a top priority, only 37% of employees drowning in rework report actually receiving it.
Gerrit Kazmaier, President of Product and Technology at Workday, argues the burden is being placed on the wrong people:
“Too many AI tools push the hard questions of trust, accuracy, and repeatability back onto individual users. Our philosophy is that AI should do the complex work under the hood so people can focus on judgment, creativity, and connection. That’s how organisations turn AI‑powered speed into durable, human‑led advantage.”
AI Value: What Separates Winners from Losers
Not every organization is trapped in this cycle. The Workday research identifies a clear dividing line between those capturing real value and those chasing the illusion of productivity.
The difference comes down to reinvestment strategy.
Struggling companies pour AI time savings back into more technology (39%) or pile on additional tasks (32%), whereas successful organizations reinvest in people. Among employees achieving strong AI outcomes, 79% have received increased skills training. These workers use recovered time for deeper analysis and strategic thinking (57%), not just clearing a longer to-do list.
As Workday’s report highlights:
“The most successful organisations don’t just deploy AI—they reinvest the time it saves into their people. By building skills, redesigning roles, and modernising how work gets done, these companies turn speed into sustained business impact.”
Four Moves for UC Leaders
Unified communications and collaboration leaders are uniquely positioned to break the AI productivity paradox. Here’s how:
Stop measuring vanity metrics. “Hours saved” means nothing if half that time disappears into correction and verification. Measure net productivity: time saved minus time spent on rework.
Close the training gap. The 30-point disconnect between leadership intent and employee experience won’t fix itself. Provide hands-on training in real work contexts, not theoretical overviews. Teach judgment frameworks for when to use AI and when to walk away.
Redesign work, not just workflows. Update job descriptions to reflect what AI can actually do. Redefine roles around judgment, creativity, and strategic thinking. Remove legacy processes that create friction with modern tools.
Choose tools that carry their weight. AI should reduce user burden, not increase it. Evaluate platforms based on built-in accuracy mechanisms, context-awareness, and transparency about limitations. If your team is spending hours fact-checking every output, the tool is pushing complexity onto users instead of handling it under the hood.
The Real AI Value Question
The AI productivity paradox forces an uncomfortable reckoning: deployment isn’t the same as value creation. Speed without quality is just waste at a faster pace.
The organizations that win won’t be those with the most AI tools. They’ll be the ones that figured out how to integrate AI into human-centered workflows—supported by training, clear roles, and honest measurement of what’s actually working.
Because right now, for most companies, the answer is: not as much as they think.
