AI And App Switching: How To Read The Harvard Study

  • by

Read the full analysis: AI And App Switching: How To Read The Harvard Study on ThorstenMeyerAI.com

TL;DR

A Harvard Business School preprint analysed 103 million application events from 1,017 office workers at eight companies in India. It found fewer switches immediately after AI use, but more switching on AI-use days overall; the observational data cannot show that AI caused either pattern.

A Harvard Business School preprint analysing 103 million application events from 1,017 workers found that people switched between apps less in the few uses immediately after using generative AI, but made about 200 more switches on AI-use days overall. The distinction matters because the widely shared claim that switching “drops after AI” describes a short sequence, not the full workday—and the study cannot show that AI caused either pattern.

The paper, Digital Fragmentation and Generative AI Use Across 103 Million Application Events, is by Sumer S. Vaid and Ashley V. Whillans of Harvard Business School. The researchers analysed records collected from November 2025 through January 2026 by Flowace, a commercial workforce analytics service used by the eight participating organisations in India. The paper says the study was not externally funded and that Flowace did not support or direct the research.

Across 62,693 person-days, workers made a mean of 908.3 switches between applications per day; the median was 814. The middle half of employees made between 547 and 1,130 switches per workday. An application remained active for an average of 27.8 seconds before a switch, while the median duration was 3.3 seconds. These measures count changes between applications, not every move between documents or browser tabs.

The AI finding depends on the time window. In the ten application uses after an AI use, workers used fewer apps, switched less, stayed about half a second longer per app and moved more predictably. Across whole days, however, workers used AI on days with roughly 200 more switches, and days with more AI minutes were more fragmented. The authors say the relationship could reflect AI changing the shape of the workday, workers using AI on already fragmented days, or both.

At a glance
reportWhen: Preprint posted July 7, 2026, revised A…
The developmentA widely shared headline said AI reduced workers’ app switching, but the underlying Harvard preprint reports a more mixed pattern: short-term switching fell after AI use while daily switching rose on days when workers used AI.

App switching and AI · one moment

What happens around one AI query?

The study zoomed in on every AI use: the ten app uses before it, and the ten after. Up close, work calms down a little. Zoom out to the whole day, and the picture changes.

10 app uses before

AI

median 7 seconds
10 app uses after

Whole AI day: about 200 more switches

1 · A workday of switchingWorkers changed apps 908 times a day on average. One app held their focus for 27.8 seconds on average, 3.3 seconds at the median.
2 · AI comes in at a busy momentPeople used AI a median of 13 times a day, for a median of 7 seconds. Fragmentation peaked in the hour of AI use.
3 · The next ten app usesFewer different apps, fewer switches, about half a second longer per app, more predictable. Real, but small.
4 · Zoom out: the whole dayOn days with AI use, the same people made about 200 more switches. Whether AI calms work or busy moments bring AI: unknown.

Source: Sumer S. Vaid, Ashley V. Whillans (Harvard Business School), “Digital Fragmentation and Generative AI Use Across 103 Million Application Events”, arXiv:2607.06681v2, preprint under review. Blocks are schematic, not to scale.

Graphic: Powered by Thorsten Meyer AI

Why the AI Pattern Changes by Window

The findings caution against treating an immediate change after AI use as evidence that AI makes an entire workday more focused. A short sequence after an AI interaction may look calmer even while the broader day contains more app changes. For workers and managers, the time window changes the interpretation.

The paper also points to daily conditions as a substantial part of the picture. The authors attribute 44.6% of variation in fragmentation to day-to-day changes within the same person, compared with 35.8% to differences between people and 19.6% to differences between organisations. That suggests a worker’s schedule and workload on a particular day may matter more than a stable label such as “focused” or “distracted.”

The study is not evidence that AI improves productivity, work quality or employee wellbeing. It measured application activity, not whether the work was better or completed faster. Its results do not support buying an AI product on the assumption that it will serve as a focus tool.

How the Workday Data Was Collected

The researchers received anonymised records from Flowace monitoring software installed on participating employees’ computers. According to the paper, the software records when the active application changes, including the application name and timing; the record can also include a window title and, where applicable, a web address. Employers chose which apps counted as “productive.”

The sample covered eight organisations in software, accounting, financial services, e-publishing, law and marketing. It included 1,017 employees, with a median of 58 observed workdays per person. The study received approval from Harvard’s human-subjects committee, and the data were provided to the researchers under a data-use agreement.

The manuscript was first posted on arXiv on July 7, 2026, and revised on August 7. It identifies itself as a preprint under peer review and warns that findings may change during that process. The paper compares observed app-stay durations with earlier attention research, but says methods differ; the comparison should be read as a broad signal, not a precise measure of how attention has changed over time.

“The manuscript is currently under review, and the findings may change based on the peer review process.”

— Sumer S. Vaid and Ashley V. Whillans, the paper’s authors

What the Records Cannot Establish

The study is observational, not an experiment. It cannot establish whether AI use led to more daily switching, whether workers turned to AI because their days were already fragmented, or whether other factors influenced both. The authors explicitly leave open the possibility that more than one explanation applies.

The sample also limits how far the findings can be generalised. It covers workers at eight organisations in India using employer-provided computer monitoring; it does not cover other countries or phone use. The study measured app activity, not the quality or value of work, and it does not show whether workers experienced the recorded switching as disruptive.

Peer review is ongoing, and the results could change. The source material also does not establish that the headline’s account on Hacker News was operated by one of the paper’s authors. The available evidence supports describing the study and its reported measurements, not treating a social post or a simplified headline as a complete account.

Peer Review and Better Focus Measures

The paper’s next research milestone is completion of peer review, which may lead to revisions or changes in how the results are presented. Until then, its conclusions should be treated as preliminary. Follow-up research would need to test cause and effect and examine whether changes in app switching correspond to changes in task performance, work quality or employee experience.

For teams considering the findings, the paper offers a reason to look at workload and schedules alongside technology use, not a prescription to deploy or remove AI. The researchers report that fragmentation varied substantially from day to day and peaked on Tuesdays, with patterns resetting after weekends and holidays. Those observations may guide questions about communication load and concentrated work time, but they do not establish a universal workplace rule.

Key Questions

Did the study find that AI reduces app switching?

Not across the whole day. Switching was lower in the ten application uses after AI use, but workers made about 200 more switches on days when they used AI.

How many workers and app events were included?

The Harvard Business School researchers analysed 103 million application events from 1,017 workers at eight organisations in India, covering November 2025 to January 2026.

Does the study prove AI causes more switching?

No. It is an observational study, not an experiment. The researchers say AI could affect the workday, workers could use AI on days that are already more fragmented, or both.

Has the study been peer reviewed?

Not yet, according to the paper. It was posted as a preprint under peer review, and its authors say the findings may change through that process.

Does the research show that AI improves productivity?

No. The researchers measured application activity and switching, not work quality, productivity or whether employees completed tasks faster.

Source: ThorstenMeyerAI.com

Leave a Reply

Your email address will not be published.