Generative AI feedback loops drive cognitive engagement and equity via the digital Hawthorne effect

Generative AI feedback loops drive cognitive engagement and equity via the digital Hawthorne effect | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Generative AI feedback loops drive cognitive engagement and equity via the digital Hawthorne effect QIN YUFENG This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9121799/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The scalable acquisition of classroom process data is chronically hindered by the trilemma of cost, privacy, and technological accessibility, particularly in low-resource environments. Here, we present a zero-hardware, privacy-preserving generative AI agent that operates via browser-based edge computing and Large Language Models (LLMs) to provide real-time formative assessment of classroom discourse. In a 16-week quasi-experimental study involving 12th-grade students (N = 98), we investigated the system’s impact on pedagogical dynamics and academic outcomes. Results demonstrate that continuous, non-intrusive AI visualization triggered a sustained "Digital Hawthorne Effect," culminating in a 376.9% increase in active classroom participation. "Furthermore, AI-quantified substantive engagement emerged as a positive protective factor, effectively buffering students against severe academic regression."Controlling for baseline abilities, the experimental group achieved significantly higher academic outcomes (partial η² = 0.063) while effectively preventing severe academic regression. Notably, the intervention exhibited a pro-equity "catch-up effect," yielding a 29% higher marginal academic benefit for low-achieving learners compared to their high-achieving peers. By replacing costly biometric surveillance with edge-processed semantic analysis, this study reframes generative AI as a low-cost environmental variable. Ultimately, our findings provide a scalable, frugal innovation framework to foster human-AI triadic symbiosis and advance global educational equity. Social science/Education Business and commerce/Information systems and information technology Physical sciences/Mathematics and computing Social science/Science technology and society Full Text Additional Declarations No competing interests reported. Supplementary Files reproducehawthorne.py reproducepaperresults.py hawthorneplotdata.xlsx plotsourcedata.xlsx README.md Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9121799","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":612927782,"identity":"bcd7592b-7e66-425c-b3f9-317a047e4dd2","order_by":0,"name":"QIN YUFENG","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA20lEQVRIiWNgGAWjYDACZiCu/GMjxw/iJBQQq+VsQ5qxZANIiwGxNp1tOJy44QCIRYwWvuO8xyQO7jhsbHx+deKHBwYM8vxiB/BrkTzMlyZx8Ey6nNmNt5slgA4znDk7Ab8Wg8M8ZtIf2KyNzW6c3QDSkmBwmwgtEgfYmBM3zzi7+QfxWg62OSdu4O/dRpwtkod5jC0OnEkzlrjBu80iwUCCsF/4zp8xvHGgAhiV/Wc33/xRYSPPL01AC8MBBhYJMEMCrFKCgHKIFuYPYAb/ASJUj4JRMApGwYgEADvoSO3EN0x/AAAAAElFTkSuQmCC","orcid":"","institution":"shenzhen middle school","correspondingAuthor":true,"prefix":"","firstName":"QIN","middleName":"","lastName":"YUFENG","suffix":""}],"badges":[],"createdAt":"2026-03-14 10:38:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9121799/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9121799/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107488681,"identity":"27078463-6544-44a8-8fa1-51b5ab239732","added_by":"auto","created_at":"2026-04-22 02:45:34","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":677708,"visible":true,"origin":"","legend":"","description":"","filename":"GenerativeAIfeedbackloopsdrivecognitiveNATURE.25599docx.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9121799/v1_covered_f99d802e-e5f1-4c4a-a209-af6d253a72a7.pdf"},{"id":105751941,"identity":"21355156-f752-46fd-8dfa-7e78cc03b771","added_by":"auto","created_at":"2026-03-30 15:51:30","extension":"py","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":10305,"visible":true,"origin":"","legend":"","description":"","filename":"reproducehawthorne.py","url":"https://assets-eu.researchsquare.com/files/rs-9121799/v1/5b64055f1d50b61205768571.py"},{"id":105577119,"identity":"c6f01d94-89ac-4170-b8cb-808f953d1c36","added_by":"auto","created_at":"2026-03-27 13:52:36","extension":"py","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":21470,"visible":true,"origin":"","legend":"","description":"","filename":"reproducepaperresults.py","url":"https://assets-eu.researchsquare.com/files/rs-9121799/v1/c28b52e47a5f35416b03b5be.py"},{"id":105577124,"identity":"ebb1d9e8-60b5-461b-826f-f75164714c00","added_by":"auto","created_at":"2026-03-27 13:52:36","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":6384,"visible":true,"origin":"","legend":"","description":"","filename":"hawthorneplotdata.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-9121799/v1/b59e38142c3b547eee46ac57.xlsx"},{"id":105577121,"identity":"6b04d155-28d0-4092-8dc9-6a25a6be0363","added_by":"auto","created_at":"2026-03-27 13:52:36","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":15065,"visible":true,"origin":"","legend":"","description":"","filename":"plotsourcedata.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-9121799/v1/d2ca9d708de3953cb0255a03.xlsx"},{"id":105577122,"identity":"380341e9-2446-470b-a307-824532895cc8","added_by":"auto","created_at":"2026-03-27 13:52:36","extension":"md","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":2037,"visible":true,"origin":"","legend":"","description":"","filename":"README.md","url":"https://assets-eu.researchsquare.com/files/rs-9121799/v1/0d8f9e3ee3738606db843490.md"}],"financialInterests":"No competing interests reported.","formattedTitle":"Generative AI feedback loops drive cognitive engagement and equity via the digital Hawthorne effect","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-9121799/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9121799/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe scalable acquisition of classroom process data is chronically hindered by the trilemma of cost, privacy, and technological accessibility, particularly in low-resource environments. Here, we present a zero-hardware, privacy-preserving generative AI agent that operates via browser-based edge computing and Large Language Models (LLMs) to provide real-time formative assessment of classroom discourse. In a 16-week quasi-experimental study involving 12th-grade students (N\u0026thinsp;=\u0026thinsp;98), we investigated the system\u0026rsquo;s impact on pedagogical dynamics and academic outcomes. Results demonstrate that continuous, non-intrusive AI visualization triggered a sustained \"Digital Hawthorne Effect,\" culminating in a 376.9% increase in active classroom participation. \"Furthermore, AI-quantified substantive engagement emerged as a positive protective factor, effectively buffering students against severe academic regression.\"Controlling for baseline abilities, the experimental group achieved significantly higher academic outcomes (partial η\u0026sup2; = 0.063) while effectively preventing severe academic regression. Notably, the intervention exhibited a pro-equity \"catch-up effect,\" yielding a 29% higher marginal academic benefit for low-achieving learners compared to their high-achieving peers. By replacing costly biometric surveillance with edge-processed semantic analysis, this study reframes generative AI as a low-cost environmental variable. Ultimately, our findings provide a scalable, frugal innovation framework to foster human-AI triadic symbiosis and advance global educational equity.\u003c/p\u003e","manuscriptTitle":"Generative AI feedback loops drive cognitive engagement and equity via the digital Hawthorne effect","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-27 13:52:31","doi":"10.21203/rs.3.rs-9121799/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"370f1f6e-b5e5-485c-9af7-efa67ee1d205","owner":[],"postedDate":"March 27th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":65220985,"name":"Social science/Education"},{"id":65220986,"name":"Business and commerce/Information systems and information technology"},{"id":65220987,"name":"Physical sciences/Mathematics and computing"},{"id":65220988,"name":"Social science/Science technology and society"}],"tags":[],"updatedAt":"2026-04-21T17:10:24+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-27 13:52:31","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9121799","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9121799","identity":"rs-9121799","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}