Fairness-Aware Deep Learning for Job Application Screening

Fairness-Aware Deep Learning for Job Application Screening | 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Research Article Fairness-Aware Deep Learning for Job Application Screening Timothy Adeyemi This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9087718/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 Automated hiring systems offer scalability, yet they often reinforce previous biases, disproportionately filtering out women, minorities, and older applicants, a significant issue as AI becomes increasingly common in recruitment decisions. A distinctive decision-level adversarial framework is introduced to tackle this problem, guaranteeing algorithmic fairness while eliminating access to sensitive features during inference. This approach employs a lightweight discriminator to deduce gender exclusively from the predictor's output probability, rather than altering latent representations, while the primary model is adversarially refined to diminish this signal, effectively decoupling hiring recommendations from demographic leakage while preserving predictive capabilities. In a dataset comprising 73,462 applicants, the suggested approach achieves an AUC of 0.875 (+0.004 compared to the baseline), an F1-score of 0.810, and a disparate effect ratio of 0.907, which surpasses the 0.8 regulatory standard for fairness. SHAP research shows that decisions are affected by job-related factors like technical skills and experience, instead of using proxy variables. Ablation studies indicate that gentle adversarial regularization enhances generalization, suggesting that fairness standards might function as efficient regularizers. These results offer a scalable, understandable, and ethically grounded approach for fair AI-driven recruitment, showing that fairness and performance can enhance concurrently rather than being mutually exclusive. Algorithmic fairness adversarial learning deep learning job application screening explainable AI bias mitigation automated hiring ethical AI 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-9087718","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":604036673,"identity":"7feaf78b-e60e-4582-bdf0-c8f887eecdeb","order_by":0,"name":"Timothy Adeyemi","email":"data:image/png;base64,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","orcid":"https://orcid.org/0009-0007-3000-8088","institution":"WeAreGenius Research Institute","correspondingAuthor":true,"prefix":"","firstName":"Timothy","middleName":"","lastName":"Adeyemi","suffix":""}],"badges":[],"createdAt":"2026-03-10 20:57:31","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":true,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":true},"doi":"10.21203/rs.3.rs-9087718/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9087718/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104780830,"identity":"6b6950c2-095c-486c-882c-9a8328bdc7cb","added_by":"auto","created_at":"2026-03-17 07:54:04","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":273827,"visible":true,"origin":"","legend":"","description":"","filename":"mrtimothyinitialsubmission1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9087718/v1_covered_49330ce3-78bc-4721-afce-c62c4620b769.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eFairness-Aware Deep Learning for Job Application Screening\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"WeAreGenius Research Institute","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":"Algorithmic fairness, adversarial learning, deep learning, job application screening, explainable AI, bias mitigation, automated hiring, ethical AI","lastPublishedDoi":"10.21203/rs.3.rs-9087718/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9087718/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAutomated hiring systems offer scalability, yet they often reinforce previous biases, disproportionately filtering out women, minorities, and older applicants, a significant issue as AI becomes increasingly common in recruitment decisions. A distinctive decision-level adversarial framework is introduced to tackle this problem, guaranteeing algorithmic fairness while eliminating access to sensitive features during inference. This approach employs a lightweight discriminator to deduce gender exclusively from the predictor's output probability, rather than altering latent representations, while the primary model is adversarially refined to diminish this signal, effectively decoupling hiring recommendations from demographic leakage while preserving predictive capabilities. In a dataset comprising 73,462 applicants, the suggested approach achieves an AUC of 0.875 (+0.004 compared to the baseline), an F1-score of 0.810, and a disparate effect ratio of 0.907, which surpasses the 0.8 regulatory standard for fairness. SHAP research shows that decisions are affected by job-related factors like technical skills and experience, instead of using proxy variables. Ablation studies indicate that gentle adversarial regularization enhances generalization, suggesting that fairness standards might function as efficient regularizers. These results offer a scalable, understandable, and ethically grounded approach for fair AI-driven recruitment, showing that fairness and performance can enhance concurrently rather than being mutually exclusive.\u003c/p\u003e","manuscriptTitle":"Fairness-Aware Deep Learning for Job Application Screening","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-12 06:31:26","doi":"10.21203/rs.3.rs-9087718/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":"d8291c1d-99e1-4d13-828e-9efeafbfb4f9","owner":[],"postedDate":"March 12th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-12T06:31:26+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-12 06:31:26","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9087718","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9087718","identity":"rs-9087718","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}