The Cognitive Fingerprint Problem: Transformer-LSTM Perplexity Geometry for Fair and Adversarially Robust AI Text Detection

The Cognitive Fingerprint Problem: Transformer-LSTM Perplexity Geometry for Fair and Adversarially Robust AI Text Detection | 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 Research Article The Cognitive Fingerprint Problem: Transformer-LSTM Perplexity Geometry for Fair and Adversarially Robust AI Text Detection Pranil Raichura This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9006491/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 As AI text detection systems become increasingly embedded in academic integrity enforcement, recent empirical audits have exposed a severe and systematic bias against English as a Second Language (ESL) writers. Standard detectors, which rely predominantly on sub-word perplexity measured against models trained on native English, frequently conflate the natural linguistic simplicity of human language learners with the statistical predictability of Large Language Models (LLMs). Furthermore, naive retraining strategies that incorporate ESL data to mitigate this bias introduce a critical new adversarial vulnerability: LLMs can be explicitly prompted to simulate ESL writing patterns---including grammatical errors and constrained vocabulary---allowing them to evade even ESL-aware detectors. The result is a scenario in which innocent ESL students are falsely accused of academic misconduct while a deliberately adversarial AI goes undetected. To resolve this dual challenge, we propose a novel feature-extraction architecture termed Geometric Fairness. Rather than measuring how "fluent" a text is on a single scale, this system maps every essay into a two-dimensional bivariate feature space defined by two orthogonal perplexity signals: (1) a Native-Expert, a pre-trained DistilGPT-2 Transformer computing sub-word perplexity against standard American English; and (2) an ESL-Expert, a custom Character-Level Long Short-Term Memory (LSTM) network trained exclusively on the ELLIPSE learner corpus, computing next-character perplexity. By contrasting sub-word probability under standard English against the sequence-level character variance of authentic learner interlanguage, our architecture geometrically separates the cognitive stochasticity of human error from the probabilistic smoothing of AI generation. On the most adversarial classification task considered---distinguishing authentic ESL student writing from AI-generated text specifically prompted to mimic ESL style---our dual-stream logistic regression classifier achieved an accuracy of 93.3%, outperforming a single-dimensional perplexity baseline of 75.0% by 18.3 percentage points. The dual-stream method achieved a Receiver Operating Characteristic Area Under the Curve (ROC-AUC) of 0.96, compared to 0.73 for the 1D ESL-only baseline. Critically, the false-positive rate---the proportion of genuine human ESL students who would be wrongly accused---was reduced from 29.0% to 9.7%, a relative reduction of 66.7%. Artificial Intelligence and Machine Learning AI Text Detection Algorithmic Fairness Bivariate Perplexity Character-Level Long Short-Term Memory Recurrent Neural Network (LSTM RNN) Interlanguage Adversarial Robustness ESL Detection Natural Language Processing (NLP) Transformers Perplexity Educational Equity LSTM Full Text Additional Declarations The authors declare no competing interests. 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-9006491","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":599055732,"identity":"5ed927f3-45c9-4183-b034-4e170c7c21d9","order_by":0,"name":"Pranil Raichura","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/ElEQVRIiWNgGAWjYLCChAIgwczABmLLgYgDDwhqMUBoMQZrSSBojQGYBGtJbAAbgkexufThZw8eGDDk87czP3vMu8cufX7Y4YdAW+zkdBuwa7HsSzM3ADrMcsZhNnNjnmfJuRtvpxkAtSQbmx3A4aQzDGYSQC0GDId52KR5DjDnbpydANJyIHEbTi3s38Ba5CFa6tMNZ6d/IKCFB2KLAUTL4QR56Rz8tlj28JQBtUgYGB5mM5Occ+C44QbpnIIDCQa4/WLOw75N8keFjYHc+cPPJN4cqJaXn52++cOHCjs5nN6HUBJIIgeQxPFoQQLyDbhVj4JRMApGwcgEAHV1Vnt7DSasAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0009-0006-6192-8488","institution":"Granite Bay High School","correspondingAuthor":true,"prefix":"","firstName":"Pranil","middleName":"","lastName":"Raichura","suffix":""}],"badges":[],"createdAt":"2026-03-02 06:43:43","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-9006491/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9006491/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106092963,"identity":"e75006e8-89d4-47b8-8f42-1c485d176a48","added_by":"auto","created_at":"2026-04-03 11:31:38","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1455230,"visible":true,"origin":"","legend":"","description":"","filename":"miniESLpaper16.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9006491/v1_covered_c86ebe03-faba-49b8-a082-bc04cc789980.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eThe Cognitive Fingerprint Problem: Transformer-LSTM Perplexity Geometry for Fair and Adversarially Robust AI Text Detection\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Granite Bay High School","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":"AI Text Detection, Algorithmic Fairness, Bivariate Perplexity, Character-Level Long Short-Term Memory Recurrent Neural Network (LSTM RNN), Interlanguage, Adversarial Robustness, ESL Detection, Natural Language Processing (NLP), Transformers, Perplexity, Educational Equity, LSTM","lastPublishedDoi":"10.21203/rs.3.rs-9006491/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9006491/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAs AI text detection systems become increasingly embedded in academic integrity enforcement, recent empirical audits have exposed a severe and systematic bias against English as a Second Language (ESL) writers. Standard detectors, which rely predominantly on sub-word perplexity measured against models trained on native English, frequently conflate the natural linguistic simplicity of human language learners with the statistical predictability of Large Language Models (LLMs). Furthermore, naive retraining strategies that incorporate ESL data to mitigate this bias introduce a critical new adversarial vulnerability: LLMs can be explicitly prompted to simulate ESL writing patterns---including grammatical errors and constrained vocabulary---allowing them to evade even ESL-aware detectors. The result is a scenario in which innocent ESL students are falsely accused of academic misconduct while a deliberately adversarial AI goes undetected.\u003c/p\u003e\n\u003cp\u003eTo resolve this dual challenge, we propose a novel feature-extraction architecture termed Geometric Fairness. Rather than measuring how \"fluent\" a text is on a single scale, this system maps every essay into a two-dimensional bivariate feature space defined by two orthogonal perplexity signals: (1) a Native-Expert, a pre-trained DistilGPT-2 Transformer computing sub-word perplexity against standard American English; and (2) an ESL-Expert, a custom Character-Level Long Short-Term Memory (LSTM) network trained exclusively on the ELLIPSE learner corpus, computing next-character perplexity. By contrasting sub-word probability under standard English against the sequence-level character variance of authentic learner interlanguage, our architecture geometrically separates the cognitive stochasticity of human error from the probabilistic smoothing of AI generation.\u003c/p\u003e\n\u003cp\u003eOn the most adversarial classification task considered---distinguishing authentic ESL student writing from AI-generated text specifically prompted to mimic ESL style---our dual-stream logistic regression classifier achieved an accuracy of 93.3%, outperforming a single-dimensional perplexity baseline of 75.0% by 18.3 percentage points. The dual-stream method achieved a Receiver Operating Characteristic Area Under the Curve (ROC-AUC) of 0.96, compared to 0.73 for the 1D ESL-only baseline. Critically, the false-positive rate---the proportion of genuine human ESL students who would be wrongly accused---was reduced from 29.0% to 9.7%, a relative reduction of 66.7%.\u003c/p\u003e","manuscriptTitle":"The Cognitive Fingerprint Problem: Transformer-LSTM Perplexity Geometry for Fair and Adversarially Robust AI Text Detection","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-03 13:47:13","doi":"10.21203/rs.3.rs-9006491/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":"d88b6006-bcfa-4ce6-8807-ad739ed92e90","owner":[],"postedDate":"March 3rd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":63744221,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2026-03-23T21:01:36+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-03 13:47:13","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9006491","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9006491","identity":"rs-9006491","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}