The key facts

  • More than 50% of executives say AI has made it harder to evaluate employee performance; 78% have changed evaluation criteria
  • 94% of executives are optimistic about AI, but 46% fear losing control over autonomous agents
  • 57% report an increase in the number of software tools instead of the expected reduction

AI has made evaluating employee performance more complicated

Companies are adopting AI tools to boost productivity, but a Nielsen survey commissioned by Monday.com of 1,700 directors and senior managers in seven countries shows it is creating new problems for executives. More than half of respondents said that after adopting AI it became harder for them to evaluate the work of their subordinates: a significant share of results arrives in finished form, and it is difficult to determine how much was done by the employee and how much by AI. Nearly half of managers reported that they now spend more time reviewing and supervising their teams' work than before.

Enthusiasm persists, but organizations can't keep up with the technology

At the same time, executives are generally positive: 94% of respondents said they are enthusiastic about improving their work with AI. Michal Miller Levi, head of marketing research and customer insights at Monday.com, compared this enthusiasm to the early days of the internet in the 2000s. According to her, the problem is that AI is developing faster than companies can adapt: 'AI works very well for them individually, but they also have another role — as managers. Their struggle stems from the fact that AI is moving faster than organizational processes.'

Evaluation criteria are changing — and evaluations themselves are increasingly done with AI

78% of respondents said they have changed their approach to evaluating employees, paying more attention to the ability to write prompts, decide which tasks to delegate to AI, and experiment with new ways of working. Meanwhile, 46% of managers use AI for decision-making, and 39% of executives use it to write performance reviews. The publication notes the risk of introducing additional biases: research by the Luxembourg Institute of Socio-Economic Research showed that, in the absence of clear objective standards, LLMs evaluate results similarly to humans, and work from early 2025 revealed bias in models against Black male candidates. When asked about the possible introduction of external bias, Levi said companies have not yet developed approaches to such problems.

'Invisible work' and a growing tool stack

The executive's workday itself is changing too: someone has to decide which tasks to hand over to AI agents, check their results, and intervene when things go wrong — the report calls this a new layer of 'invisible work.' Software has also become more complex: 57% of managers say there are more tools in organizations than before, about 51% note increased interdependence between tools, and only 29% describe their AI infrastructure as fully optimized and centralized. According to Levi, businesses expected AI to help shrink the software stack, but instead the number of applications, integrations, and 'shadow' use of AI outside approved systems is growing: 'More than 50% say they are using more tools than last year, and it is not stopping.'

Concerns about autonomous agents

As AI agents capable of acting independently are deployed, the need for oversight grows: 58% of managers expect to supervise more AI agents than people over the next year, and 46% fear losing control over their actions. 35% of respondents fear being held responsible for decisions made by AI, since LLMs remain largely unpredictable. 'They don't want their agents to be fully autonomous,' Levi noted, adding that executives want humans to be involved in key tasks and decisions. Overall, as TechCrunch summarizes, adopting AI is becoming not only an opportunity to boost productivity but also a management and control challenge.