Self-Aware Language Models: A Taxonomy and Evaluation of Epistemic Uncertainty and Hallucination Mitigation

Self-Aware Language Models: A Taxonomy and Evaluation of Epistemic Uncertainty and Hallucination Mitigation | 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 Self-Aware Language Models: A Taxonomy and Evaluation of Epistemic Uncertainty and Hallucination Mitigation Anjikya Tiwari, Vibhuti Gupta This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8589677/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 Artificial intelligence is rapidly getting ubiquitous, getting intertwined with people's everyday lives. As the adoption grows, so does the need to verify large language models (LLMs) for correctness. The inability of LLMs to recognize their own knowledge gaps remains a fundamental limitation. While prior research address subsets of these mechanisms, they do not analyze them explicitly as manifestations of epistemic self-awareness nor research their interaction as a unified capability. In this review we investigate knowledge gap awareness, an emerging field that seeks to enable LLMs with epistemic self-awareness to detect, present and respond to the absence of knowledge. We synthesize research across five key mechanisms: (i) reflective prompting, (ii) uncertainty quantification, (iii) selective prediction and abstention, (iv) retrieval-based verification, and (v)confidence calibration. Drawing on 51 curated papers using PRISMA methodology, we provide a comprehensive taxonomy, evaluate current benchmarks, and analyze practical applications in medicine, science, education, law, and collaborative AI. We also outline open challenges and limitations in modeling epistemic context-drive knowledge gap monitoring in LLMs and its applications in read world. Hallucination detection responsible AI trustworthy AI epistemic uncer- tainty confidence calibration retrieval-augmented generation (RAG) AI safety and reliability Full Text Additional Declarations No competing interests reported. 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-8589677","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":576298397,"identity":"6644c622-98f1-4783-908c-e9ffeb8d7c33","order_by":0,"name":"Anjikya Tiwari","email":"","orcid":"","institution":"Microsoft (United States)","correspondingAuthor":false,"prefix":"","firstName":"Anjikya","middleName":"","lastName":"Tiwari","suffix":""},{"id":576298398,"identity":"764cc55c-e2ed-4d89-8a43-281a155addfb","order_by":1,"name":"Vibhuti Gupta","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwUlEQVRIiWNgGAWjYDACZh42MM0PYgMxYwOxWiQkG5iJ1cIA1WJwgFgtBsd5jz382nanzvhG/tHNBQw2shsOENJymC/dWLbtmYTZjWS22zMY0owJapFs5jGTlmw7DNHCw3A4kXgtxjPAWv4T1sLPzGMm+RGoxUACrOUAMVr40qQZzj2TnHHmsdntGQbJxjMJaWHjP3tM8kfZHX7+9sRntwsq7GT7CGkBAWage6BMAyKUgwDjD7iWUTAKRsEoGAVYAAAkjkC0pOCAggAAAABJRU5ErkJggg==","orcid":"","institution":"The University of Texas Medical Branch at Galveston","correspondingAuthor":true,"prefix":"","firstName":"Vibhuti","middleName":"","lastName":"Gupta","suffix":""}],"badges":[],"createdAt":"2026-01-13 09:08:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8589677/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8589677/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107047812,"identity":"1f11a6bd-eccf-475d-b07d-1fb962792057","added_by":"auto","created_at":"2026-04-16 07:43:28","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":780051,"visible":true,"origin":"","legend":"","description":"","filename":"MindtheGap.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8589677/v1_covered_b60c2dab-e9bc-4c72-b886-7a582d3b512e.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Self-Aware Language Models: A Taxonomy and Evaluation of Epistemic Uncertainty and Hallucination Mitigation","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"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":"Hallucination detection, responsible AI, trustworthy AI, epistemic uncer- tainty, confidence calibration, retrieval-augmented generation (RAG), AI safety and reliability","lastPublishedDoi":"10.21203/rs.3.rs-8589677/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8589677/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eArtificial intelligence is rapidly getting ubiquitous, getting intertwined with people's everyday lives. As the adoption grows, so does the need to verify large language models (LLMs) for correctness. The inability of LLMs to recognize their own knowledge gaps remains a fundamental limitation. While prior research address subsets of these mechanisms, they do not analyze them explicitly as manifestations of epistemic self-awareness nor research their interaction as a unified capability. In this review we investigate knowledge gap awareness, an emerging field that seeks to enable LLMs with epistemic self-awareness to detect, present and respond to the absence of knowledge. We synthesize research across five key mechanisms: (i) reflective prompting, (ii) uncertainty quantification, (iii) selective prediction and abstention, (iv) retrieval-based verification, and (v)confidence calibration. Drawing on 51 curated papers using PRISMA methodology, we provide a comprehensive taxonomy, evaluate current benchmarks, and analyze practical applications in medicine, science, education, law, and collaborative AI. We also outline open challenges and limitations in modeling epistemic context-drive knowledge gap monitoring in LLMs and its applications in read world.\u003c/p\u003e","manuscriptTitle":"Self-Aware Language Models: A Taxonomy and Evaluation of Epistemic Uncertainty and Hallucination Mitigation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-27 17:22:59","doi":"10.21203/rs.3.rs-8589677/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":"cc9c69fb-daac-4fdc-b5b3-9e369aa350dd","owner":[],"postedDate":"January 27th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-16T07:41:11+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-27 17:22:59","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8589677","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8589677","identity":"rs-8589677","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}