An analysis of more than 9,000 employees suggests heavier AI usage is associated with longer workdays and fragmented attention, challenging assumptions about automation and efficiency.
Artificial intelligence is supposed to help employees accomplish more in less time. New workplace research suggests that promise is proving considerably more complicated.
A study released by ActivTrak’s Productivity Lab found that employees using AI recorded workday spans of 8.4 to 9.1 hours, compared with 7.6 hours for non-users—a difference of up to 90 minutes daily.
The analysis covered 9,140 knowledge workers across 467 enterprises between April 1 and September 25, 2026.
The findings raise an important question for companies investing in workplace automation: are employees actually gaining time, or simply taking on more work?
Heavy AI Users Experience More Fragmented Work
ActivTrak examined digital activity patterns rather than relying on employees’ self-reported estimates.
The findings showed that heavy AI users, defined as employees spending more than 45 minutes across at least five daily sessions, experienced shorter uninterrupted focus periods than light users.
Their average focus session lasted 37.8 minutes, compared with 48.2 minutes among light users—a difference of approximately 22%.
Heavy users also switched applications around 15% more frequently than employees who did not use AI.
However, the researchers identified differences depending on how AI tools were integrated into existing software.
Employees using embedded AI averaged 0.6 application switches per hour, compared with between 1.7 and 4.2 among users of browser-based AI tools.
Software Integration May Matter More Than Adoption
According to CIO’s analysis of the findings, published November 3, the evidence complicates conventional assumptions about AI productivity.
Earlier ActivTrak research, covering 163,638 employees across 1,111 organizations, found that overall workdays had shortened slightly.
The apparent discrepancy highlights differences in sample sizes, observation periods and employee groups.
It also exposes the limitations of measuring productivity primarily through time spent working or using AI.
Gabriela Mauch, ActivTrak’s chief customer officer and head of its Productivity Lab, emphasized that organizations should evaluate actual productivity and business performance rather than treating AI adoption as an achievement by itself.
More Output Does Not Necessarily Mean Less Pressure
The new findings also reveal that heavy AI users spend approximately 18 additional minutes working after hours compared with non-users.
Their proportion of after-hours activity is similar, but their longer working days increase the absolute amount of time involved.
Meanwhile, a separate Gartner report published October 26 found that 77% of board members expected AI to deliver productivity improvements and cost savings.
Yet only 3% of organizations exceeded their cost-saving goals through productivity improvements.
Gartner senior director analyst CV Viverito also reported that 38% of companies were over budget on AI, while 85% planned to increase spending the following year.
The Productivity Paradox Has Important Limitations
The ActivTrak analysis is observational. It does not demonstrate that AI caused longer working hours.
Employees handling demanding assignments may be more likely to adopt AI, while different job responsibilities could explain variations in application switching.
As TechRadar’s November 5 coverage emphasized, the findings challenge companies to distinguish meaningful improvements from increased software activity.
For founders and product leaders, that means evaluating completed work, accuracy, employee workload and customer outcomes alongside adoption metrics.
The emerging challenge is ensuring AI improves how people work rather than simply increasing how much work organizations expect them to complete.
