🔍 Read the full analysis: A Closer Look At AI, App Switching, And The Harvard Study on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
A Harvard Business School preprint analyzed 103 million application events from 1,017 workers at eight companies in India. It found fewer app switches in the ten application uses after AI use, but about 200 more switches on days when workers used AI; the observational data cannot show that AI caused either pattern.
A Harvard Business School preprint examining 103 million application events found that workers switched between apps less in the ten uses immediately after using generative AI, but made about 200 more switches on AI-use days overall. The findings complicate a headline that spread on November 18 claiming that app switching drops after AI: the study is observational and does not establish that AI caused either pattern.
Researchers Sumer S. Vaid and Ashley V. Whillans analyzed second-by-second activity records from 1,017 office workers at eight organizations in India, covering November 2025 through January 2026. The dataset included 103 million application events and 62,693 person-days. Across the records, employees switched applications a mean of 908.3 times per day; the median was 814 switches per workday.
The paper reports two different patterns around AI. In the ten application uses after an AI use, workers used fewer apps, switched less, and stayed about half a second longer in each app. Their activity also became more predictable. But on days when workers used AI, they made about 200 additional switches per day, and days with more AI minutes were more fragmented. These results describe associations in the records, not a tested effect of using AI.
The study, titled Digital Fragmentation and Generative AI Use Across 103 Million Application Events, is a preprint under peer review. The researchers received anonymized records from Flowace, a commercial workforce-analytics service used by the participating companies. The paper says employers determined which applications counted as productive. It did not measure whether the work improved, whether workers felt more focused, or the quality of their output.
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.
- 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
AI Use and Workday Fragmentation
The findings matter because app switching is being used to support claims about whether AI makes office work more focused. The study does not support a simple conclusion that AI reduces interruptions: the immediate sequence after AI use looks less fragmented, while the overall day-level association points in the opposite direction. Those differences are a reason to distinguish a short window around a tool interaction from the broader shape of a workday.
The paper also suggests that day-to-day conditions may matter more than fixed differences between workers or organizations. Its analysis attributes 44.6% of variation in fragmentation to changes within the same person across days, compared with 35.8% to differences between people and 19.6% to differences between organizations. The authors report fragmentation peaking on Tuesdays and resetting after weekends and holidays. These patterns may help frame workplace questions, but they do not show which interventions would improve focus or results.
For employers, the results are not evidence that buying an AI product will make staff more attentive. The source account advises treating fragmentation as a feature of the day, while recognizing that the paper has not tested measures such as protected focus periods. For workers, the headline figure describes recorded app changes, not a direct measure of distraction, productivity, or the value of their work.
Products Worth Considering

Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite

Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Bluetooth 5.3, USB-C Charging, IPX4, Black

Kindle Paperwhite Signature Edition 32GB (2024 Model) – Metallic Black
As an affiliate, we earn on qualifying purchases.
How the App Records Were Collected
The research covers employees at eight Indian organizations in software, accounting, financial services, e-publishing, law, and marketing. Each worker contributed a median of 58 workdays. Flowace’s software recorded an event when the active application changed, with information such as the app name and the start and end time. The paper says the researchers received the data anonymized under a data-use agreement and that the study protocol was approved by Harvard’s human-subjects committee.
The paper was first posted to arXiv on July 7, 2026, and revised on August 7. Its PDF identifies it as a preprint and warns that findings may change through peer review. The source account says a November 18 Hacker News post helped spread the “after AI, switching drops” framing; it could not confirm who operated the account that posted it. The headline’s first claim—that workers averaged about 900 app switches per day—closely reflects the study’s reported mean of 908.3. The AI framing needs the day-level result as well as the short-window result.
The authors state that the study had no external funding and that Flowace provided no funding or commercial support and had no role in study design, analysis, or interpretation. The paper also discloses that Vaid holds equity in a private company unrelated to Flowace that uses AI to create enterprise digital twins. Those disclosures describe the reported funding and interests; they do not resolve the study’s limits of measurement or scope.
“The manuscript is currently under review, and the findings may change based on the peer review process.”
— Sumer S. Vaid and Ashley V. Whillans, in the preprint
What the Study Cannot Establish
The data cannot determine whether AI use leads to more fragmented days, whether workers choose AI on days that are already more fragmented or consolidated, or whether both processes occur. The researchers describe the direction of the relationship as unknown; the study did not randomly assign AI use or otherwise test cause and effect.
The sample also limits how widely the results can be applied. It covers employees at eight organizations in India, not workers in other countries or people using phones. Because the records came from employer monitoring software, they capture application activity rather than every part of a worker’s attention or experience. The paper does not establish whether app switching harmed concentration, changed productivity, or improved or reduced work quality. Peer review is still underway, and the published findings could change.
Peer Review and Further Evidence
The next formal step is peer review of the preprint, which may lead to revisions or a published version. The source material does not give a review timetable. Readers should check for a revised paper and updated methods or results before treating the current estimates as settled.
To establish whether AI itself changes switching, future research would need to separate AI use from the demands of the day—for example, by comparing workers or work periods under a design capable of testing causal effects. Until then, the reported figures show patterns in one monitored sample, not proof that an AI tool improves focus. The current evidence supports a narrower conclusion: app switching was lower immediately after AI use but higher on AI-use days overall, and the reason remains unresolved.
Key Questions
Did the Harvard 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. The study does not show that AI caused either result.
What does the figure of 908 app switches mean?
It is the study’s mean number of application changes per day across its recorded data. The median was 814 switches per workday. The figure counts changes to a different application; it is not a direct measure of distraction or productivity.
Who was included in the research?
The analysis covered 1,017 office workers at eight organizations in India, across fields including software, accounting, financial services, publishing, law, and marketing. Records ran from November 2025 through January 2026.
Is the study peer reviewed?
No. The paper is a preprint under peer review. Its authors caution that findings may change during that process.
Does the study show that AI improves productivity?
No. The records measured application activity, not work quality or productivity. The study also cannot establish whether AI use caused the observed switching patterns.
Source: ThorstenMeyerAI.com
