Does workplace AI save employees time, or add work and interruptions? Two ActivTrak reports point in different directions, underscoring how difficult it is to measure AI’s effect on productivity.
The company’s latest study tracked behavioral data from 9,140 knowledge workers at 467 enterprises between April 1 and September 25, 2026. Heavy AI users worked between 8.42 and 9.08 hours a day, compared with 7.58 hours for employees who did not use AI. Their focus sessions were also 22% shorter than those of light users: 37.8 minutes versus 48.2 minutes. Heavy users switched between apps 15% more often.
Those findings differ from ActivTrak’s earlier research, which analyzed 443 million work hours from 163,638 employees at 1,111 organizations between 2023 and 2025. That report found that average workdays fell by roughly nine minutes, or 2%, with AI. It also reported a 19-minute increase in daily productive time, reaching six hours and 36 minutes, while productive sessions grew by 13%.
The studies also differ in scale and duration, making direct comparisons difficult. ActivTrak’s earlier report found that focus efficiency fell to a three-year low of 60% during the study period, and average focused session time declined 9% from 2023 to 2025. Multitasking increased by 12%, or more than 10 minutes.
In the latest research, employees using AI tools embedded in other software switched apps 0.6 times an hour. Those using web-based AI switched between 1.7 and 4.2 times an hour—up to four times as often as employees who did not use AI. ActivTrak describes the extra effort of moving between applications as a “swivel-chair tax.” Light AI users also spent a larger share of active work time in core job-specific apps than heavy users: 81.4%, compared with 78%.
Sarah Altemus, manager of the ActivTrak Productivity Lab, said more frequent AI use does not necessarily mean more focused work. She also cautioned that any time saved could be offset if managers assign employees additional responsibilities without setting clear workload limits.
Sanchit Vir Gogia, chief analyst at Greyhound Research, said ActivTrak can show how work moves across applications but noted that the company sells the tools used to make those measurements. He also pointed out that tracking systems may miss AI use on personal devices, work completed by background agents or APIs, and activity that does not produce a desktop session. App switching, he added, is easier to observe than shifts in attention within an application.
Gogia said organizations evaluating AI should look beyond whether an initial task takes less time and establish whose work benefited—and who took on the effort behind that improvement.
Sources: Computerworld.