AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Does AI Change How Often Workers Switch Apps? on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get the latest gadgets delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

A Harvard Business School preprint analyzing 103 million application events found workers switched apps more on days they used AI, but switched less in the ten app uses following an AI use. The observational data cannot show whether AI caused either pattern, and the study did not measure work quality.

A Harvard Business School preprint analyzing 103 million application events from 1,017 office workers found that employees switched between apps more on days they used generative AI, while making fewer switches in the ten app uses after an AI interaction, a workplace pattern also explored in workers’ views on AI and daily work. The two patterns complicate claims that AI simply reduces workplace distraction: the study is observational, under peer review and cannot establish whether AI changes switching behavior or is used on days that are already more fragmented.

The paper, “Digital Fragmentation and Generative AI Use Across 103 Million Application Events,” was written by Sumer S. Vaid and Ashley V. Whillans of Harvard Business School. Researchers examined records from 1,017 employees at eight organizations in India, gathered from November 2025 through January 2026 using Flowace workforce-analytics software. The sample included workers in software, accounting, financial services, e-publishing, law and marketing.

Across 62,693 person-days, the researchers counted 56.9 million switches between applications. Workers averaged 908.3 switches per day; the median was 814. The paper defines a switch as a change to a different active application, not a move between files within the same app. It reports an average app-use period of 27.8 seconds, with a median of 3.3 seconds. Those figures describe measured application activity, not necessarily attention or task completion.

The AI findings differ by time scale. In the ten app uses after an AI use, employees used fewer applications, switched less often, spent about half a second longer in each app and moved in a more predictable sequence. Yet on days with any AI use, employees made about 200 more switches per day on average; days with more AI minutes were also more fragmented. These are associations in logged activity, not results from an experiment.

At a glance
reportWhen: Preprint first posted July 7, 2026, rev…
The developmentA large Harvard Business School preprint reports that AI use is associated with more app switching across a workday but less switching immediately after an AI interaction.

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. 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. 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. 3 · The next ten app usesFewer different apps, fewer switches, about half a second longer per app, more predictable. Real, but small.
  4. 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

AI Use Shows Two Switching Patterns

The findings matter because businesses often present AI tools as a way to make work more efficient or reduce routine task-switching. This study does not support a simple claim that AI makes the whole workday less fragmented. Instead, its results point to a distinction between what happens immediately after an AI interaction and the broader pattern on a day when AI is used.

That distinction also limits what employers can conclude from monitoring data. A day with more app switching might reflect interruptions, a heavier workload, or a different mix of tasks; the study does not determine which. It also did not measure whether employees produced better work, finished tasks faster or experienced more stress. The authors report that day-to-day changes within the same worker accounted for 44.6% of variation in fragmentation, compared with 35.8% tied to differences between workers and 19.6% between organizations. That suggests daily conditions deserve attention, but does not identify which workplace changes would improve outcomes.

For readers deciding whether AI is a focus aid, the practical takeaway is limited: this paper alone is not evidence that buying an AI tool will reduce interruptions. Nor does the higher daily switch count prove that AI causes distraction. Both causal claims go beyond the study design.

Vitamix 5-Speed Immersion Blender, 625-Watt, Stainless Steel/Black

Vitamix 5-Speed Immersion Blender, 625-Watt, Stainless Steel/Black

Apple AirPods Pro (2nd Generation) Wireless Earbuds with USB-C MagSafe Charging Case

Apple AirPods Pro (2nd Generation) Wireless Earbuds with USB-C MagSafe Charging Case

As an affiliate, we earn on qualifying purchases.

How Researchers Tracked App Activity

The research used data collected by Flowace, a commercial workforce-analytics service already used by the participating organizations. Its software records application changes on employees’ computers, including app names and use times. Employers, rather than the researchers, decided which applications counted as productive. The records were anonymized before being provided to the research team under a data-use agreement, and the study protocol received approval from Harvard’s human-subjects committee.

The paper was first posted on arXiv on July 7, 2026, and revised on August 7. It identifies itself as a preprint under review and cautions that findings may change during peer review. The authors state that the research had no external funding and that Flowace provided no commercial support or role in study design, analysis or interpretation. One author disclosed equity in a separate company that uses AI to create enterprise digital twins.

The headline that circulated on November 18 emphasized that workers switched apps about 900 times a day and suggested switching drops after AI. The count is close to the paper’s 908.3 average, but the AI result needs both time scales to be understood: less switching immediately after AI use coexisted with more switching on AI-use days overall.

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

— The authors, in the preprint

What the Data Cannot Establish

The study cannot determine whether AI use leads to more switching, whether workers reach for AI when their day is already busy, or whether both happen. The researchers explicitly leave the direction of the relationship unresolved. Because the records cover workers at eight organizations in India, the results may not apply to other countries, industries or people working primarily on phones.

Application logs also cannot show why a person changed apps or whether the change interrupted focused work. The reported app-use duration is not a direct measure of attention. The study did not assess work quality, output, employee experience or productivity, so it cannot show whether the measured switching helped or harmed performance. Peer review is still pending, and the findings may change.

Peer Review and Broader Evidence

The paper’s next formal milestone is completion of peer review; the source material does not give a publication date or a schedule for further results. Readers should check for a reviewed version before treating the findings as settled. Further research would need to test whether AI changes switching behavior, rather than merely tracking AI use alongside app activity, and should include varied workplaces and direct measures of work outcomes.

Until then, the evidence supports a narrow conclusion: in this dataset, AI-use days had more app switching overall, while the sequence of activity immediately after AI use was less fragmented. What causes those patterns, and whether they matter for performance, remains unanswered.

Key Questions

Did workers switch apps more or less when using AI?

Both patterns appeared. Workers switched less in the ten app uses after an AI interaction, but made about 200 more switches on AI-use days overall, on average.

Does the study prove AI causes more app switching?

No. The research is observational and cannot establish whether AI caused the switching patterns, workers used AI because their day was already fragmented, or both.

How many workers and organizations were included?

The study analyzed activity from 1,017 employees at eight organizations in India, covering November 2025 through January 2026.

Is the Harvard study peer-reviewed?

Not yet, according to the source material. The paper is an arXiv preprint under review, and its authors say the findings may change during peer review.

Does the paper show that AI improves productivity?

No. It measures application activity and switching, not work quality, output or productivity. It cannot establish whether the observed patterns helped or harmed workers’ performance.

Source: ThorstenMeyerAI.com

You May Also Like

Wizards Of The Coast Surges In Global Coverage

Search interest and media mentions of Wizards Of The Coast have sharply increased, with 128 mentions in recent coverage, signaling rising global attention.

Breaking Down The Hype: GLM-5.3-Flash As A Cheap AI Engine

An in-depth analysis of GLM-5.3-Flash, a 320-billion-parameter multimodal model released by Z.ai, highlighting its cost-efficiency and potential for AI agents.

How To Assess The Seven-Neuron Risk In An AI Agent

Microsoft researchers report that Lazarus can insert trigger-based payloads into AI agents by editing fewer than ten neurons, without retraining.

A Museum Tour Of AI: The 12 Questions Everyone Has

Exploring common questions about AI through a virtual museum, clarifying what AI is, how it works, and what remains uncertain.