AI intensifies work — and only workstyle data can spot it

The productivity dashboard has rarely looked better. Output per head is rising. Messages move faster, and more of the work now spills across more hours of the day. The people generating those numbers are quietly fraying, and the dashboard has no way to tell the difference.

That gap is the subject of an eight-month study at a 200-person technology firm, where researchers expected generative AI to lighten the load and found the opposite. AI did not give workers their time back. It made them work faster, widen the scope of what they treated as their job, and push work into more of the day, usually without anyone asking them to.¹² The freed minutes were not banked as breathing room. They were reinvested in more work.

Intensification took three forms. People absorbed tasks that used to belong to someone else. The natural pauses in a day dissolved, as prompts fired during lunch, before meetings, and late at night. And workers kept several threads alive at once, sometimes running multiple AI agents in parallel.¹² None of this registers as a problem on a volume metric. All of it shows up later as attrition.

The dashboard reads intensity as engagement

All of this leaves a measurable trace. A before-and-after analysis of 443 million hours of activity across 163,638 workers found that once AI tools entered the workflow, email time rose 104% and chat 145%, while daily focus time fell about 9%, roughly 23 minutes a day.³ Weekend work climbed too, with Sunday hours up 58%.³ If AI were genuinely substituting for effort, total work time would fall. Instead it rose and splintered.

The trouble is that every one of those movements looks like good news on a conventional dashboard. More messages, more systems touched, more output per hour: a volume metric was built to count exactly these things and to read all of them as productivity. It cannot register whether the pace behind them is sustainable, because it was never asked to.³¹ This is a measurement problem dressed as a productivity story.

Recovery is the next casualty. Because AI makes any task trivial to start and to continue, work seeps into the moments that used to function as stopping points. Microsoft's analysis of trillions of workplace signals shows the same boundary loss at population scale: 40% of employees check email before 6 a.m., and meetings after 8 p.m. are up 16% year on year.⁴

The workday no longer has edges, and a dashboard that counts activity will only ever see more of it to applaud.

The cost arrives as overload before it arrives as hours

The first symptom of unsustainable pace is not a longer timesheet. It is cognitive overload. A new strain is emerging from the work of supervising, checking, and stitching together AI output, what some have begun calling 'AI brain fry'. Around 14% of nearly 1,500 surveyed workers reported significant AI-related mental fatigue, and close to half of all employees say their work feels chaotic and fragmented.⁵⁴ Fragmentation and decision fatigue, not just long days, are what tip people toward the exit.

What makes the pattern so hard to catch is that it feels good at first. The extra effort is self-directed and often genuinely exciting, so what was once above and beyond slowly becomes the baseline. Greater capability produces greater output, which raises expectations, which pressures further expansion. The loop stays invisible in output numbers because output is precisely the thing that keeps improving.¹²

It would be easy to read this as an argument against AI. It is not. The enthusiasm is real, the short-run gains are real, and the researchers behind the original study stress both.⁵² That is the trap rather than the rebuttal: voluntary, exciting effort is exactly what makes the reset invisible, because nobody resents a pace they chose. The work is not to brake adoption but to build a feedback loop around it, pairing behavioural signals with an explicit practice of sequencing work, protecting focus, and defending recovery.¹

This is where workstyle analytics earns its place. It is not another activity counter; it is the one instrument that can separate genuine gain from unsustainable acceleration — pace change measured against a team's own baseline, after-hours drift, and the steady narrowing of focus windows. The data already sits in most organisations. The question is whether anyone is reading it before the intensity reads as turnover. As AI makes every team faster, which leaders will be able to tell a team that is thriving from one that is merely accelerating?

Footnotes

1. Ye, X. M., & Ranganathan, A. (2026, February). AI doesn't reduce work—it intensifies it. Harvard Business Review. https://hbr.org/2026/02/ai-doesnt-reduce-work-it-intensifies-it ↩ ↩² ↩³ ↩⁴ ↩⁵

2. Counts, L. (2026, February 18). AI promised to free up workers' time. UC Berkeley Haas researchers found the opposite. Berkeley Haas News. https://newsroom.haas.berkeley.edu/ai-promised-to-free-up-workers-time-uc-berkeley-haas-researchers-found-the-opposite ↩ ↩² ↩³ ↩⁴

3. Moss, W. (2026, March 18). AI isn't killing jobs—it may be intensifying work instead. Forbes. https://www.forbes.com/sites/wesmoss/2026/03/18/ai-isnt-killing-jobs-it-may-be-intensifying-work-instead ↩ ↩² ↩³

4. Microsoft. (2025, June 17). Breaking down the infinite workday. Microsoft Work Trend Index Special Report. https://www.microsoft.com/en-us/worklab/work-trend-index/breaking-down-infinite-workday ↩ ↩²

5. HRD Connect. (2026, March 16). Is AI helping burnout or quietly making it worse? HRD Connect. https://www.hrdconnect.com/2026/03/16/is-ai-helping-burnout-or-quietly-making-it-worse ↩ ↩²