AI Was Sold as a Productivity Tool. For Managers, It's a Measurement Crisis

More than half of senior leaders say AI adoption has made it harder to evaluate their employees' work, according to a Nielsen survey of 1,700 directors and senior managers across seven countries, commissioned by work-management company Monday.com. That is the core paradox of enterprise AI in 2025: the tools deliver, but nobody can cleanly say who or what did the delivering.

By Olivia Hart

The mechanics of the problem are blunt. Much of employees' output now arrives polished in the manager's inbox, and managers cannot reliably separate the human contribution from the machine's. Nearly half of managers say they now spend more time reviewing and guiding their teams' output than they did before AI arrived. The technology that promised to save time is, at the layer of supervision, consuming it.

Is the enthusiasm real, or just premature?

The survey's respondents are not skeptics. 94% of managers polled said they were excited to improve work using AI, and many said they know where to deploy it for the most value. Michal Miller Levi, head of market research and customer insights at Monday.com, compared the mood to the early internet era of the 2000s — people see a similar kind of "magic" in AI.

But Levi, speaking to TechCrunch, drew a sharp line between individual benefit and organizational readiness. "AI works very well for [managers] individually, but they also have a different role as managers. They need to drive the organization as well […] Their struggle is [due to] the fact that AI is just moving faster than the organizational processes," Levi said.

That admission matters. The vendor that paid for the survey is conceding that its customers' processes lag the technology they are buying. The excitement, in other words, is running ahead of the infrastructure.

How do you judge work you can't attribute?

Managers are already rewriting the rules. 78% of respondents said they have changed how they evaluate employees since adopting AI. The new criteria are process skills rather than raw output: how well staff write prompts, which tasks they choose to delegate to AI, and how willing they are to experiment with new ways of working.

The fairness question, however, cuts both ways. The survey found that:

• 46% of managers use AI to help inform their decisions
• 39% of leaders use AI to draft performance reviews

So some managers cannot tell how much of an employee's work came from AI — while themselves using AI to judge that work. The loop is circular, and nobody in the data claims to have closed it.

The bias risk is documented, not hypothetical. LLMs are, at their core, prediction machines that repeat human patterns embedded in their training data. Research from the Luxembourg Institute of Socio-Economic Research earlier this year found that "in the absence of clear objective standards and when individuals are rated independently," LLMs tend to rate performance similarly to humans. A separate research paper in early 2025 found AI models can be biased against Black male candidates and "demonstrate significant biases, varying in terms of the directions and magnitudes across different social groups."

When asked whether AI weighing in on performance could import outside bias into reviews, Levi said companies have yet to work out how to handle such issues. That gap — acknowledged, unresolved — sits at the center of the entire adoption wave.

Why didn't the software stack shrink?

AI is also redrawing the manager's working day. Someone must decide which tasks to hand to AI agents, check what those agents produce, and intervene when something fails. The report frames this as a new layer of "invisible labor" — supervision that never appears in any productivity metric.

The software estate itself is growing, not consolidating:

• 57% of managers say their organizations now have more tools in circulation than before
• About 51% say their tools have grown more interdependent and clustered since adopting AI
• Only 29% describe their AI infrastructure as fully optimized and centralized

Levi said businesses had expected the opposite. "Managers completed 2025 with the feeling that they will consolidate tools, actually; that they'll reduce the number of software. But what happened is that the number of tools that they're using increased," Levi said. "More than 50% are saying that they're using more tools now than they did last year, and it's not stopping […]."

Add shadow AI — employees adopting tools outside approved company systems — and the control problem compounds rather than resolves.

Who is accountable when agents act alone?

The urgency is rising because AI agents increasingly act on their own. The survey found 58% of managers expect to oversee more AI agents than human employees within the next year — a striking inversion of the span-of-control model that has defined management for a century. Yet 46% worry about losing control over those agents' actions, and 35% fear being held responsible for decisions made by AI, given that LLMs remain largely unpredictable.

"They don't want their agents to be fully autonomous," Levi said, explaining that managers want people involved and consulted by AI on crucial tasks and decisions. That is a rational hedge, not Luddism: the same managers racing to deploy agents are the ones reporting they cannot yet govern them.

So what?

The report's own findings show AI adoption becoming as much a governance problem as a productivity play. Until companies build the processes and infrastructure to attribute work, audit AI-assisted reviews, and contain sprawling tool stacks, managers will keep absorbing the cost of the gap between what the technology can do and what their organizations can manage.