AI promises productivity, but for managers, it’s mostly a headache for now

Companies are giving employees AI tools to marque them much productive, but the tools look to beryllium creating caller problems for their managers.

More than fractional of elder leaders accidental find it harder to measure employees’ enactment aft adopting AI, according to a Nielsen survey of 1,700 directors and elder managers crossed 7 countries, commissioned by the work-management institution Monday.com. The crushed is that overmuch of employees’ output present arrives polished successful the manager’s inbox, making it hard to archer however overmuch came from the worker and however overmuch from AI. As a result, astir fractional of managers accidental they walk much clip reviewing and guiding their teams’ output than they utilized to.

That’s not to accidental managers are unhappy astir AI. Quite the opposite, successful fact: 94% of the managers polled said they were excited to amended enactment utilizing AI, and galore are assured they cognize wherever to usage it to get the astir value. Michal Miller Levi, caput of marketplace probe and lawsuit insights astatine Monday.com, compared the enthusiasm to the aboriginal days of the net successful the 2000s, saying radical spot a akin benignant of “magic” to AI.

But the problem, she told TechCrunch, is simply that AI is processing faster than companies tin adapt. “AI works precise good for [managers] individually, but they besides person a antithetic relation arsenic managers. They request to thrust the enactment arsenic good […] Their conflict is [due to] the information that AI is conscionable moving faster than the organizational processes,” Levi said.

Managers are besides reconsidering however they justice their teams: 78% of respondents said they person changed however they measure employees aft adopting AI, placing greater value connected however good unit tin make prompts, determine which tasks to delegate to AI, and experimentation with caller ways of working.

But however reasonably tin employees beryllium judged erstwhile managers can’t easy archer however overmuch of their enactment was produced by AI? And what if the managers themselves are progressively relying connected AI to marque those judgments? The survey recovered that 46% of managers usage AI to assistance pass their decisions, and 39% of leaders are utilizing AI to draught show reviews.

That signifier runs the hazard of introducing adjacent much bias than accustomed into the reappraisal process, fixed that LLMs are, astatine their core, prediction machines, and thin to repetition quality patterns recovered successful their grooming data. Research from the Luxembourg Institute of Socio-Economic Research earlier this twelvemonth found that “in the lack of wide nonsubjective standards and erstwhile individuals are rated independently,” LLMs thin to complaint show likewise to humans. And a research paper successful aboriginal 2025 recovered that AI models tin beryllium biased against achromatic antheral candidates, and “demonstrate important biases, varying successful presumption of the directions and magnitudes crossed antithetic societal groups.”

When I asked Levi whether utilizing AI to measurement successful connected people’s show could present extracurricular bias into the process, she said companies person yet to enactment retired however to grip specified issues.

AI is besides changing what a manager’s time looks like. Someone has to determine which tasks to manus implicit to AI agents, cheque what they produce, and measurement successful erstwhile thing goes wrong. The study suggests that this supervision could magnitude to a caller furniture of “invisible labor.”

The bundle itself is becoming harder to manage. Fifty-seven percent of managers accidental their organizations present person much tools successful circulation than before, and astir 51% said their tools had grown much interdependent and clustered since adopting AI. Only 29% described their AI infrastructure arsenic afloat optimized and centralized.

Levi said businesses had expected AI to assistance shrink their bundle stacks. Instead, they’re dealing with much apps, integrations, and shadiness AI usage arsenic employees follow tools extracurricular their companies’ approved systems.

“Managers completed 2025 with the feeling that they volition consolidate tools, actually; that they’ll trim the fig of software. But what happened is that the fig of tools that they’re utilizing increased,” Levi said. “More than 50% are saying that they’re utilizing much tools present than they did past year, and it’s not stopping […].”

This request for power seems to beryllium increasing urgent arsenic companies deploy AI agents that tin enactment connected their own. The survey recovered that 58% of managers expect to oversee much AI agents than quality employees wrong the adjacent year, and 46% interest astir losing power implicit those agents’ actions.

There’s besides rather a spot of fearfulness implicit being held liable for decisions made by AI (35% of respondents said so), arsenic LLMs are inactive mostly unpredictable. “They don’t privation their agents to beryllium afloat autonomous,” Levi said, explaining that managers privation radical to beryllium progressive and consulted by AI erstwhile it comes to important tasks and decisions.

At heart, the report’s findings amusement that AI adoption is progressively becoming arsenic overmuch a governance occupation arsenic it is an accidental to rise productivity. Without due processes and infrastructure to woody with it, managers look to beryllium scrambling to support things chugging.