Companies may be deploying artificial intelligence faster than their workforces are learning how to supervise it. An IBM Institute for Business Value study summarized by eeNews Europe found a sharp difference between the skills employers prioritize and those employees believe matter most.
The research, conducted with Oxford Economics between April and June 2026, surveyed 1,500 chief human resources officers and 8,800 full-time employees across multiple countries and industries. Seventy-one percent of the HR leaders identified the ability to supervise, validate and override AI output as the most essential skill. Only 29% of employees ranked judgment as important.
AI can create hidden work
The findings complicate the idea that automation simply removes tasks. Eighty percent of the HR leaders said AI adoption creates “invisible” work, including checking recommendations, correcting errors, adding context and handling exceptions. Forty-two percent of employees said AI either increases their workload or produces work that goes unrecognized.
Accountability is also uneven. Forty-three percent of employees said they are blamed when something goes wrong with AI, while 41% of HR leaders believe workers may not feel safe challenging or overriding an automated output. Sixty percent of employees worry that using AI could erode skills they already have.
Clear roles correlate with better results
IBM found that organizations defining which workflows are human-led, AI-assisted or AI-executed reported an 18% reduction in risk and a 20% improvement in quality. Those figures describe associations within IBM’s study, not a controlled proof that role definitions alone caused the improvement.
Critical thinking ranked highly on both sides of the survey, cited by 49% of employees and 57% of HR leaders. The practical lesson is that AI training cannot stop at teaching people how to produce prompts. Workers also need authority, time and clear responsibility to question an output, correct it and decide when automation should not control the task.
