I cracked open the ActivTrak report late, and the numbers read like a slow leak: people who use AI are staying at their desks longer. I felt that small jolt you get when a trend you’ve heard whispered becomes a headline. You start to notice the evenings stretching, not shrinking.
I’ve chased workplace studies for years, and I want to give you the straight read: the narrative that AI will hand us a three‑day weekend is meeting a stubborn countercurrent. Bill Gates and NVIDIA CEO Jensen Huang have floated the idea of shorter workweeks as AI scales; Senator Bernie Sanders has pushed for a 32‑hour week to protect workers. But the fresh data from ActivTrak and researchers at UC Berkeley’s Haas School pushes a different, sharper question at us: who benefits when AI speeds up the engine?
At a glance: employees using AI are logging longer days — here’s the math
ActivTrak analyzed activity from more than 9,000 employees across 467 companies between April and September and found a clear gap. AI users averaged between 8.4 and 9.1 hours a day, while non‑users averaged about 7.6 hours. That’s up to 90 extra minutes of work daily for people who lean on AI.
The study defines heavy AI users as people who spend more than 45 minutes across at least five sessions a day using AI tools. Those workers were also doing an extra 18 minutes of work after hours compared with non‑users, switching apps more frequently and seeing shorter stretches of uninterrupted focus.
Does AI make employees work longer?
Yes—at least according to recent measurements. ActivTrak’s data and a separate eight‑month ethnographic study at a 200‑person tech firm by UC Berkeley’s Haas School both show that AI often increases pace and scope of work. Gabriela Mauch of ActivTrak put it plainly: AI can help you complete tasks faster, but it appears to stretch the workday, fragment attention, and ratchet expectations upward.
At my desk: faster tools, thicker to‑do lists
I watched this pattern become routine in teams I’ve covered: a shortcut shows up, and the list grows. The Haas researchers observed that AI didn’t free up time so much as multiply what people were expected to handle. Employees worked faster, widened their task scope, and let work bleed into more hours—often without being asked.
Xingqi Maggie Ye, one of the study’s authors, warned that what starts as thrilling productivity can harden into a new baseline of performance. The effect is subtle: extra hours and responsibilities become the norm, a reset of expectations you may not notice until it’s standard.
Will AI lead to a shorter workweek?
That’s the high‑stakes bet backed by figures like Gates and Huang and advocated politically by Sanders, who supports a 32‑hour week with no loss in pay. The evidence today, however, shows a messy middle: employers may see productivity gains on paper, while workers shoulder denser, more intense days. If gains are not redistributed, a shorter workweek remains an aspiration, not a reality.
At the team level: context and control matter
You can’t read the data without a second question: are workers choosing more work, or is AI driving that choice? There are plausible mechanisms in both directions. Some employees who already push long hours adopt AI to be even more efficient. Others take on more tasks because AI lowers the friction to try them. And then there’s the time spent sorting imperfect AI output—what some call “AI slop”—which can eat minutes into hours.
Think of it like a rubber band being stretched: managers feel productivity gains at first, but tension on the other end rises until someone snaps—or resets the rules.
At the company level: measurement is an advantage and a risk
Companies that measure digital activity—ActivTrak included—can spot gains quickly and tilt incentives toward speed. That’s useful for leaders, but it can be a trap for workers when faster becomes the new baseline without pay or boundary changes. The Haas team described parallel AI‑assisted workflows that made days denser and cognitively heavier.
When a workplace treats speed as the main metric, employees end up adding lanes to their own highway of tasks, which can feel efficient but makes everything harder to sustain.
How much longer do AI users work?
According to ActivTrak’s dataset, AI users’ average workday ranges from 8.4 to 9.1 hours versus 7.6 hours for non‑users—up to about 90 minutes more. Heavy users add roughly 18 minutes of after‑hours time and report more frequent app switching and shorter focus sessions.
At policy level: the redistribution question
Senator Sanders argues for a 32‑hour week so gains from automation and AI don’t accrue only to corporate leaders. That frames the broader debate: if AI elevates output, who captures the value? Corporations may pocket efficiency in revenue, and workers may pay with longer, more intense days unless policy or bargaining changes the split.
You can see why this matters: productivity gains that don’t reach salaries or time‑off policies erode the promise many hoped AI would deliver.
What managers and workers should watch now: who decides the goals, how success is measured, and whether time saved is converted into more work or more life. I’ll be watching the next reports and the bargaining fights that follow—but what do you think is a fair trade when AI makes a workday longer?
