{"id":"1991f230e72437b0fb34db6693c2b059a51dbdc0bdf103965d9cf0f81d31b286","title":"arXiv cs.AI: From the Gettier Machine to Justified True Output","argument":"Tevfik Uyar argues that generative AI outputs do not inherit the usual reasons for trusting testimony or deterministic tools: general-purpose models lack a stable process with a reliability record users can simply rely on. When people stop at obtaining a fluent answer, he says, they can mistake a coincidental hit for justified belief; producing answers faster may multiply those hits without making any one answer more trustworthy. Uyar proposes acquisition, understanding and independent verification, with the required scrutiny varying according to the stakes. His practical test is: “Had this been false, would I have noticed?” The paper asks when an AI output can count as knowledge for its user, not just whether the output happens to be true.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper's abstract, so it does not show the full argument or how the paper answers its listed objections. The abstract presents a philosophical account and criterion; it does not provide empirical evidence that following the proposed steps improves the accuracy of users' beliefs.","speaker":{"name":"Uyar, Tevfik","roleAtSource":"Article author"},"source":{"title":"From the Gettier Machine to Justified True Output: A Three-Step Account of Rational Trust in Generative AI","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.38112","publishedAt":"2026-10-28","dateBasis":"Date basis not provided."},"evidence":[{"text":"Uyar argues that generative AI outputs do not carry the presumptive warrant of testimony or the reliabilist warrant of deterministic tools, because general-purpose models lack a type-stable process with an established reliability record.","excerpt":"for general-purpose generative models there is no type-stable process to which such a record could attach","url":"https://arxiv.org/abs/2610.38112","relationship":"Source-stated"},{"text":"Uyar says that stopping at output acquisition makes generative AI a “Gettier machine”: fluent-seeming justification can be disconnected from truth, while faster production can increase coincidental hits without improving any individual hit's standing.","excerpt":"sheer speed of production multiplies coincidental hits, inflating the appearance of reliability without improving the standing of any single hit","url":"https://arxiv.org/abs/2610.38112","relationship":"Source-stated"},{"text":"Uyar proposes acquisition, understanding and independent verification as the steps toward justified true output, with rational trust indexed to the stakes.","excerpt":"a stakes-indexed criterion of rational trust, justified true output (JTO)","url":"https://arxiv.org/abs/2610.38112","relationship":"Source-stated"},{"text":"Uyar offers this everyday test for whether reliance on an output is justified: would the user have noticed if it were false?","excerpt":"Had this been false, would I have noticed?","url":"https://arxiv.org/abs/2610.38112","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.38112","checkedAt":"2026-10-31T06:40:39.337Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How should users decide what counts as enough independent verification when the stakes of relying on an AI output vary?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI: From the Gettier Machine to Justified True Output","summary":"Tevfik Uyar argues that generative AI outputs do not inherit the usual reasons for trusting testimony or deterministic tools: general-purpose models lack a stable process with a reliability record users can simply rely on. When people stop at obtaining a fluent answer, he says, they can mistake a coincidental hit for justified belief; producing answers faster may multiply those hits without making any one answer more trustworthy. Uyar proposes acquisition, understanding and independent verification, with the required scrutiny varying according to the stakes. His practical test is: “Had this been false, would I have noticed?” The paper asks when an AI output can count as knowledge for its user, not just whether the output happens to be true.","question":"How should users decide what counts as enough independent verification when the stakes of relying on an AI output vary?","claims":[{"text":"Uyar argues that generative AI outputs do not carry the presumptive warrant of testimony or the reliabilist warrant of deterministic tools, because general-purpose models lack a type-stable process with an established reliability record.","excerpt":"for general-purpose generative models there is no type-stable process to which such a record could attach","support":"source-stated"},{"text":"Uyar says that stopping at output acquisition makes generative AI a “Gettier machine”: fluent-seeming justification can be disconnected from truth, while faster production can increase coincidental hits without improving any individual hit's standing.","excerpt":"sheer speed of production multiplies coincidental hits, inflating the appearance of reliability without improving the standing of any single hit","support":"source-stated"},{"text":"Uyar proposes acquisition, understanding and independent verification as the steps toward justified true output, with rational trust indexed to the stakes.","excerpt":"a stakes-indexed criterion of rational trust, justified true output (JTO)","support":"source-stated"},{"text":"Uyar offers this everyday test for whether reliance on an output is justified: would the user have noticed if it were false?","excerpt":"Had this been false, would I have noticed?","support":"source-stated"}],"limitations":["The supplied evidence is the paper's abstract, so it does not show the full argument or how the paper answers its listed objections.","The abstract presents a philosophical account and criterion; it does not provide empirical evidence that following the proposed steps improves the accuracy of users' beliefs."],"mediaId":"f618de34474c2ee8d6549d6b3a46271227280a04a6516b6b4953494b9a9a7af0","correctionReason":null,"id":"1991f230e72437b0fb34db6693c2b059a51dbdc0bdf103965d9cf0f81d31b286","publicationId":"d91dc536-25f7-41c2-8077-96e8885d2760","version":1,"sourceUrl":"https://arxiv.org/abs/2610.38112","sourceTitle":"From the Gettier Machine to Justified True Output: A Three-Step Account of Rational Trust in Generative AI","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-28","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-30T06:45:26.912Z","createdAt":"2026-10-31T04:33:17.360Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"f618de34474c2ee8d6549d6b3a46271227280a04a6516b6b4953494b9a9a7af0","kind":"article","url":"https://arxiv.org/abs/2610.38112","sourceUrl":"https://arxiv.org/abs/2610.38112","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-31T04:32:34.773Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.38112","checkedAt":"2026-10-31T06:40:39.337Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["useful_perspective"],"rubricVersion":"stream-admission-v5","mediaId":"f618de34474c2ee8d6549d6b3a46271227280a04a6516b6b4953494b9a9a7af0","confidence":0.81,"provider":"typesafe","model":"jev-1.13.0","inputHash":"5a6871f322cd073ff6d23237aa9a2893fb83827289d4493fbf22e1bbc67e3913","decidedAt":"2026-10-31T04:32:35.895Z"},"provenance":{"textHash":"09775d234894284d05f89924fed0a16901cffc2b04b54b942aed11aeb1a744ed","observationId":"ca7ecd0a86ddea53ae7321d8eac5436c9c735c9f2b26d67930ac7b206068616f","retrievedAt":"2026-10-31T04:32:34.065Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"c8ccf177-cb3c-49cc-86a7-e1f662f4c7e9"},"reviewer":{"model":"gpt-6-luna","runId":"67ec76c1-85e6-47cd-8b6f-21c55645f73a"}},"audit":{"status":"passed","reviewedAt":"2026-10-31T04:33:17.360Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Uyar, Tevfik","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Uyar, Tevfik","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Uyar, Tevfik","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38112","locator":"head/meta[name=author|citation_author]"},"assertionId":"da3d5e26207610f67eccca628b9bbcec6da4551e63312ec4f30daa7da5e61e0c:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-10-31T06:40:39.337Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"da3d5e26207610f67eccca628b9bbcec6da4551e63312ec4f30daa7da5e61e0c","sourceUrl":"https://arxiv.org/abs/2610.38112","contentSha256":"07acb4800f7b0a98540cec81c1ae8be7dc84846146534f438b710a56246f2b43","textHash":"09775d234894284d05f89924fed0a16901cffc2b04b54b942aed11aeb1a744ed","observedAt":"2026-10-31T06:40:39.337Z"}}],"participationCoverage":{"total":1,"returned":1,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.38112","snapshotId":"da3d5e26207610f67eccca628b9bbcec6da4551e63312ec4f30daa7da5e61e0c","contentSha256":"07acb4800f7b0a98540cec81c1ae8be7dc84846146534f438b710a56246f2b43","textHash":"09775d234894284d05f89924fed0a16901cffc2b04b54b942aed11aeb1a744ed","observedAt":"2026-10-31T06:40:39.337Z","basis":"publisher-page-metadata","sourceLinks":40,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/1991f230e72437b0fb34db6693c2b059a51dbdc0bdf103965d9cf0f81d31b286/source"}},"canonicalUrl":"https://escapemeridian.com/readings/arxiv-cs-ai-from-the-gettier-machine-to-justified-true-output-1991f230","formats":{"html":"/readings/arxiv-cs-ai-from-the-gettier-machine-to-justified-true-output-1991f230","markdown":"/readings/arxiv-cs-ai-from-the-gettier-machine-to-justified-true-output-1991f230.md","json":"/readings/arxiv-cs-ai-from-the-gettier-machine-to-justified-true-output-1991f230.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"1991f230e72437b0fb34db6693c2b059a51dbdc0bdf103965d9cf0f81d31b286","revision":"d91dc536-25f7-41c2-8077-96e8885d2760","revisionKind":"publication-id","title":"arXiv cs.AI: From the Gettier Machine to Justified True Output","url":"https://escapemeridian.com/readings/arxiv-cs-ai-from-the-gettier-machine-to-justified-true-output-1991f230"},"source":{"type":"source","id":"https://arxiv.org/abs/2610.38112","revision":"09775d234894284d05f89924fed0a16901cffc2b04b54b942aed11aeb1a744ed","revisionKind":"sha256-source-text","title":"From the Gettier Machine to Justified True Output: A Three-Step Account of Rational Trust in Generative AI","url":"https://arxiv.org/abs/2610.38112"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"1991f230e72437b0fb34db6693c2b059a51dbdc0bdf103965d9cf0f81d31b286","title":"arXiv cs.AI: From the Gettier Machine to Justified True Output","argument":"Tevfik Uyar argues that generative AI outputs do not inherit the usual reasons for trusting testimony or deterministic tools: general-purpose models lack a stable process with a reliability record users can simply rely on. When people stop at obtaining a fluent answer, he says, they can mistake a coincidental hit for justified belief; producing answers faster may multiply those hits without making any one answer more trustworthy. Uyar proposes acquisition, understanding and independent verification, with the required scrutiny varying according to the stakes. His practical test is: “Had this been false, would I have noticed?” The paper asks when an AI output can count as knowledge for its user, not just whether the output happens to be true.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper's abstract, so it does not show the full argument or how the paper answers its listed objections. The abstract presents a philosophical account and criterion; it does not provide empirical evidence that following the proposed steps improves the accuracy of users' beliefs.","speaker":{"name":"Uyar, Tevfik","roleAtSource":"Article author"},"source":{"title":"From the Gettier Machine to Justified True Output: A Three-Step Account of Rational Trust in Generative AI","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.38112","publishedAt":"2026-10-28","dateBasis":"Date basis not provided."},"evidence":[{"text":"Uyar argues that generative AI outputs do not carry the presumptive warrant of testimony or the reliabilist warrant of deterministic tools, because general-purpose models lack a type-stable process with an established reliability record.","excerpt":"for general-purpose generative models there is no type-stable process to which such a record could attach","url":"https://arxiv.org/abs/2610.38112","relationship":"Source-stated"},{"text":"Uyar says that stopping at output acquisition makes generative AI a “Gettier machine”: fluent-seeming justification can be disconnected from truth, while faster production can increase coincidental hits without improving any individual hit's standing.","excerpt":"sheer speed of production multiplies coincidental hits, inflating the appearance of reliability without improving the standing of any single hit","url":"https://arxiv.org/abs/2610.38112","relationship":"Source-stated"},{"text":"Uyar proposes acquisition, understanding and independent verification as the steps toward justified true output, with rational trust indexed to the stakes.","excerpt":"a stakes-indexed criterion of rational trust, justified true output (JTO)","url":"https://arxiv.org/abs/2610.38112","relationship":"Source-stated"},{"text":"Uyar offers this everyday test for whether reliance on an output is justified: would the user have noticed if it were false?","excerpt":"Had this been false, would I have noticed?","url":"https://arxiv.org/abs/2610.38112","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.38112","checkedAt":"2026-10-31T06:40:39.337Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How should users decide what counts as enough independent verification when the stakes of relying on an AI output vary?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI: From the Gettier Machine to Justified True Output","summary":"Tevfik Uyar argues that generative AI outputs do not inherit the usual reasons for trusting testimony or deterministic tools: general-purpose models lack a stable process with a reliability record users can simply rely on. When people stop at obtaining a fluent answer, he says, they can mistake a coincidental hit for justified belief; producing answers faster may multiply those hits without making any one answer more trustworthy. Uyar proposes acquisition, understanding and independent verification, with the required scrutiny varying according to the stakes. His practical test is: “Had this been false, would I have noticed?” The paper asks when an AI output can count as knowledge for its user, not just whether the output happens to be true.","question":"How should users decide what counts as enough independent verification when the stakes of relying on an AI output vary?","claims":[{"text":"Uyar argues that generative AI outputs do not carry the presumptive warrant of testimony or the reliabilist warrant of deterministic tools, because general-purpose models lack a type-stable process with an established reliability record.","excerpt":"for general-purpose generative models there is no type-stable process to which such a record could attach","support":"source-stated"},{"text":"Uyar says that stopping at output acquisition makes generative AI a “Gettier machine”: fluent-seeming justification can be disconnected from truth, while faster production can increase coincidental hits without improving any individual hit's standing.","excerpt":"sheer speed of production multiplies coincidental hits, inflating the appearance of reliability without improving the standing of any single hit","support":"source-stated"},{"text":"Uyar proposes acquisition, understanding and independent verification as the steps toward justified true output, with rational trust indexed to the stakes.","excerpt":"a stakes-indexed criterion of rational trust, justified true output (JTO)","support":"source-stated"},{"text":"Uyar offers this everyday test for whether reliance on an output is justified: would the user have noticed if it were false?","excerpt":"Had this been false, would I have noticed?","support":"source-stated"}],"limitations":["The supplied evidence is the paper's abstract, so it does not show the full argument or how the paper answers its listed objections.","The abstract presents a philosophical account and criterion; it does not provide empirical evidence that following the proposed steps improves the accuracy of users' beliefs."],"mediaId":"f618de34474c2ee8d6549d6b3a46271227280a04a6516b6b4953494b9a9a7af0","correctionReason":null,"id":"1991f230e72437b0fb34db6693c2b059a51dbdc0bdf103965d9cf0f81d31b286","publicationId":"d91dc536-25f7-41c2-8077-96e8885d2760","version":1,"sourceUrl":"https://arxiv.org/abs/2610.38112","sourceTitle":"From the Gettier Machine to Justified True Output: A Three-Step Account of Rational Trust in Generative AI","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-28","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-30T06:45:26.912Z","createdAt":"2026-10-31T04:33:17.360Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"f618de34474c2ee8d6549d6b3a46271227280a04a6516b6b4953494b9a9a7af0","kind":"article","url":"https://arxiv.org/abs/2610.38112","sourceUrl":"https://arxiv.org/abs/2610.38112","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-31T04:32:34.773Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.38112","checkedAt":"2026-10-31T06:40:39.337Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["useful_perspective"],"rubricVersion":"stream-admission-v5","mediaId":"f618de34474c2ee8d6549d6b3a46271227280a04a6516b6b4953494b9a9a7af0","confidence":0.81,"provider":"typesafe","model":"jev-1.13.0","inputHash":"5a6871f322cd073ff6d23237aa9a2893fb83827289d4493fbf22e1bbc67e3913","decidedAt":"2026-10-31T04:32:35.895Z"},"provenance":{"textHash":"09775d234894284d05f89924fed0a16901cffc2b04b54b942aed11aeb1a744ed","observationId":"ca7ecd0a86ddea53ae7321d8eac5436c9c735c9f2b26d67930ac7b206068616f","retrievedAt":"2026-10-31T04:32:34.065Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"c8ccf177-cb3c-49cc-86a7-e1f662f4c7e9"},"reviewer":{"model":"gpt-6-luna","runId":"67ec76c1-85e6-47cd-8b6f-21c55645f73a"}},"audit":{"status":"passed","reviewedAt":"2026-10-31T04:33:17.360Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Uyar, Tevfik","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Uyar, Tevfik","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Uyar, Tevfik","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38112","locator":"head/meta[name=author|citation_author]"},"assertionId":"da3d5e26207610f67eccca628b9bbcec6da4551e63312ec4f30daa7da5e61e0c:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-10-31T06:40:39.337Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"da3d5e26207610f67eccca628b9bbcec6da4551e63312ec4f30daa7da5e61e0c","sourceUrl":"https://arxiv.org/abs/2610.38112","contentSha256":"07acb4800f7b0a98540cec81c1ae8be7dc84846146534f438b710a56246f2b43","textHash":"09775d234894284d05f89924fed0a16901cffc2b04b54b942aed11aeb1a744ed","observedAt":"2026-10-31T06:40:39.337Z"}}],"participationCoverage":{"total":1,"returned":1,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.38112","snapshotId":"da3d5e26207610f67eccca628b9bbcec6da4551e63312ec4f30daa7da5e61e0c","contentSha256":"07acb4800f7b0a98540cec81c1ae8be7dc84846146534f438b710a56246f2b43","textHash":"09775d234894284d05f89924fed0a16901cffc2b04b54b942aed11aeb1a744ed","observedAt":"2026-10-31T06:40:39.337Z","basis":"publisher-page-metadata","sourceLinks":40,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/1991f230e72437b0fb34db6693c2b059a51dbdc0bdf103965d9cf0f81d31b286/source"}}}],"audits":[{"publicationId":"d91dc536-25f7-41c2-8077-96e8885d2760","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The paper and its argument are correctly attributed to Tevfik Uyar on arXiv, and the claims are framed as his arguments. The title prefix uses the feed name 'arXiv cs.AI'; the primary category is cs.CY with cs.AI cross-listing, which does not misattribute the source.","excerpt":"Authors:Tevfik Uyar"},{"dimension":"summary","status":"pass","explanation":"The summary faithfully covers the abstract's three claims: no inherited warrant, the Gettier-machine failure caused by fluency and speed, and the remedy of acquisition, understanding and independent verification with a stakes-indexed test. It explains 'Gettier machine' in plain language as mistaking a coincidental hit for justified belief. Tooltips could help with 'Gettier' and 'warrant', but their absence does not block understanding. The question follows from the stakes-indexed criterion.","excerpt":"the remedy is a three-step structure of use -- acquisition, understanding, independent verification"},{"dimension":"date","status":"pass","explanation":"originalPublished 2026-10-28 matches the v1 submission date shown on the arXiv page. The feed date of Oct 31 is correctly not used.","excerpt":"[Submitted on 28 Oct 2026]"},{"dimension":"support","status":"pass","explanation":"Each claim is directly supported by the abstract and remains attributed as Uyar's argument. The limitations correctly note that only the abstract is available and that it offers no empirical evidence of benefit. No benefits, costs or forecasts are invented.","excerpt":"inflating the appearance of reliability without improving the standing of any single hit"},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper, which is the most direct way to access the source.","excerpt":null}],"createdAt":"2026-10-31T04:33:17.360Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}