{"id":"94e3c4bddc89872e084bb3a9b03a506e31724fa2ec69c0318d38ea0ee2564835","title":"Antonio Collante and coauthors on the impact of LLMs on code review","argument":"Antonio Collante and coauthors examine GPT-assisted GitHub pull request (PR) workflows, asking whether assistance is associated with faster code review and shorter waits for acceptance. They analyzed 25,473 PRs from 9,254 GitHub projects, identifying GPT-assisted PRs with a semi-automated method they say reached 95% labeling accuracy. The authors report that GPT-assisted PRs had a median resolution time of 9 hours, compared with 23 hours for non-assisted PRs. They also report 33% less review time and an 87% reduction in waiting time before acceptance. In a separate sample of 300 GPT-assisted PRs, they say developers mainly used GPT for code optimization, bug fixing and documentation updates. The paper’s practical point is that GPT assistance may affect different stages of review, not just the act of reviewing code. The findings give software teams concrete workflow measures to examine. Escape Meridian interpretation: the abstract describes a dataset comparison and a heuristic for identifying assisted PRs; it does not establish that GPT alone caused the reported differences.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The abstract reports a semi-automated identification method with 95% labeling accuracy; it does not describe in detail how misidentified PRs might affect the comparisons. The abstract does not establish that GPT alone caused the reported time differences.","speaker":{"name":"Collante, Antonio, Abedu, Samuel, Khatoonabadi, SayedHassan, Abdellatif, Ahmad, Alor, Ebube, Shihab, Emad","roleAtSource":"Article author"},"source":{"title":"The Impact of Large Language Models (LLMs) on Code Review Process","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2508.11034","publishedAt":"2025-08-14","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors analyzed 25,473 pull requests from 9,254 GitHub projects and used a semi-automated heuristic to identify GPT-assisted pull requests, reporting 95% labeling accuracy.","excerpt":"We curated a dataset of 25,473 PRs from 9,254 GitHub projects and identified GPT-assisted PRs using a semi-automated heuristic approach that combines keyword-based detection, regular expression filtering, and manual verification until achieving 95% labeling accuracy.","url":"https://arxiv.org/abs/2508.11034","relationship":"Source-stated"},{"text":"The authors report median resolution time of 9 hours for GPT-assisted PRs and 23 hours for non-assisted PRs, a reduction of more than 60%.","excerpt":"GPT-assisted PRs reduced median resolution time by more than 60% (9 hours compared to 23 hours for non-assisted PRs).","url":"https://arxiv.org/abs/2508.11034","relationship":"Source-stated"},{"text":"The authors report that GPT use was associated with 33% less review time and 87% less waiting time before acceptance.","excerpt":"We discovered that utilizing GPT can reduce the review time by 33% and the waiting time before acceptance by 87%.","url":"https://arxiv.org/abs/2508.11034","relationship":"Source-stated"},{"text":"In a sample of 300 GPT-assisted PRs, the authors say developers mainly used GPT for code optimization, bug fixing and documentation updates.","excerpt":"developers predominantly use GPT for code optimization (60%), bug fixing (26%), and documentation updates (12%).","url":"https://arxiv.org/abs/2508.11034","relationship":"Source-stated"},{"text":"Because the abstract describes comparisons between identified GPT-assisted and non-assisted PRs, it does not establish that GPT alone caused the reported time differences.","excerpt":"to evaluate differences between GPT-assisted and non-assisted PRs, both at the overall resolution level and across distinct review phases.","url":"https://arxiv.org/abs/2508.11034","relationship":"Escape Meridian interpretation"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2508.11034","checkedAt":"2026-11-13T10:20:08.246Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Would the reported time differences hold if GPT-assisted PRs were identified or compared using a different method?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Antonio Collante and coauthors on the impact of LLMs on code review","summary":"Antonio Collante and coauthors examine GPT-assisted GitHub pull request (PR) workflows, asking whether assistance is associated with faster code review and shorter waits for acceptance. They analyzed 25,473 PRs from 9,254 GitHub projects, identifying GPT-assisted PRs with a semi-automated method they say reached 95% labeling accuracy. The authors report that GPT-assisted PRs had a median resolution time of 9 hours, compared with 23 hours for non-assisted PRs. They also report 33% less review time and an 87% reduction in waiting time before acceptance. In a separate sample of 300 GPT-assisted PRs, they say developers mainly used GPT for code optimization, bug fixing and documentation updates. The paper’s practical point is that GPT assistance may affect different stages of review, not just the act of reviewing code. The findings give software teams concrete workflow measures to examine. Escape Meridian interpretation: the abstract describes a dataset comparison and a heuristic for identifying assisted PRs; it does not establish that GPT alone caused the reported differences.","question":"Would the reported time differences hold if GPT-assisted PRs were identified or compared using a different method?","claims":[{"text":"The authors analyzed 25,473 pull requests from 9,254 GitHub projects and used a semi-automated heuristic to identify GPT-assisted pull requests, reporting 95% labeling accuracy.","excerpt":"We curated a dataset of 25,473 PRs from 9,254 GitHub projects and identified GPT-assisted PRs using a semi-automated heuristic approach that combines keyword-based detection, regular expression filtering, and manual verification until achieving 95% labeling accuracy.","support":"source-stated"},{"text":"The authors report median resolution time of 9 hours for GPT-assisted PRs and 23 hours for non-assisted PRs, a reduction of more than 60%.","excerpt":"GPT-assisted PRs reduced median resolution time by more than 60% (9 hours compared to 23 hours for non-assisted PRs).","support":"source-stated"},{"text":"The authors report that GPT use was associated with 33% less review time and 87% less waiting time before acceptance.","excerpt":"We discovered that utilizing GPT can reduce the review time by 33% and the waiting time before acceptance by 87%.","support":"source-stated"},{"text":"In a sample of 300 GPT-assisted PRs, the authors say developers mainly used GPT for code optimization, bug fixing and documentation updates.","excerpt":"developers predominantly use GPT for code optimization (60%), bug fixing (26%), and documentation updates (12%).","support":"source-stated"},{"text":"Because the abstract describes comparisons between identified GPT-assisted and non-assisted PRs, it does not establish that GPT alone caused the reported time differences.","excerpt":"to evaluate differences between GPT-assisted and non-assisted PRs, both at the overall resolution level and across distinct review phases.","support":"interpretation"}],"limitations":["The abstract reports a semi-automated identification method with 95% labeling accuracy; it does not describe in detail how misidentified PRs might affect the comparisons.","The abstract does not establish that GPT alone caused the reported time differences."],"mediaId":"6bfd02fd08ab63389c774e86325df7d1c0b8e643893839a69ecd112f844434f2","correctionReason":null,"id":"94e3c4bddc89872e084bb3a9b03a506e31724fa2ec69c0318d38ea0ee2564835","publicationId":"1f14ffde-964c-4397-8f12-94e7b839eb02","version":1,"sourceUrl":"https://arxiv.org/abs/2508.11034","sourceTitle":"The Impact of Large Language Models (LLMs) on Code Review Process","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2025-08-14","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-13T07:43:23.424Z","createdAt":"2026-11-13T09:16:46.707Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"6bfd02fd08ab63389c774e86325df7d1c0b8e643893839a69ecd112f844434f2","kind":"article","url":"https://arxiv.org/abs/2508.11034","sourceUrl":"https://arxiv.org/abs/2508.11034","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-13T09:04:04.766Z","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/2508.11034","checkedAt":"2026-11-13T10:20:08.246Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v6","mediaId":"6bfd02fd08ab63389c774e86325df7d1c0b8e643893839a69ecd112f844434f2","confidence":0.93,"provider":"typesafe","model":"jev-1.13.0","inputHash":"ef03eedb85cf44d8635f5865a4d9b4210cb716e38a31a4da10c76e2ef4993940","decidedAt":"2026-11-13T09:04:07.440Z"},"provenance":{"textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observationId":"c457ff23a7a9dbb84e68937d89dadad71ed8e38ff1077a76cd07a64d25c1bee0","retrievedAt":"2026-11-13T09:04:04.465Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"db3a8883-b34d-455e-849c-48e167bee9ab"},"reviewer":{"model":"gpt-6-luna","runId":"2320aeca-a17a-44f9-8372-86ddade3c845"}},"audit":{"status":"passed","reviewedAt":"2026-11-13T09:16:46.707Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Collante, Antonio","url":null,"kind":"unknown"},{"name":"Abedu, Samuel","url":null,"kind":"unknown"},{"name":"Khatoonabadi, SayedHassan","url":null,"kind":"unknown"},{"name":"Abdellatif, Ahmad","url":null,"kind":"unknown"},{"name":"Alor, Ebube","url":null,"kind":"unknown"},{"name":"Shihab, Emad","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Collante, Antonio","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Collante, Antonio","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Abedu, Samuel","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Abedu, Samuel","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Khatoonabadi, SayedHassan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Khatoonabadi, SayedHassan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Abdellatif, Ahmad","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Abdellatif, Ahmad","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Alor, Ebube","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Alor, Ebube","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Shihab, Emad","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Shihab, Emad","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}}],"participationCoverage":{"total":6,"returned":6,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2508.11034","snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z","basis":"publisher-page-metadata","sourceLinks":48,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/94e3c4bddc89872e084bb3a9b03a506e31724fa2ec69c0318d38ea0ee2564835/source"}},"canonicalUrl":"https://escapemeridian.com/readings/antonio-collante-and-coauthors-on-the-impact-of-llms-on-code-94e3c4bd","formats":{"html":"/readings/antonio-collante-and-coauthors-on-the-impact-of-llms-on-code-94e3c4bd","markdown":"/readings/antonio-collante-and-coauthors-on-the-impact-of-llms-on-code-94e3c4bd.md","json":"/readings/antonio-collante-and-coauthors-on-the-impact-of-llms-on-code-94e3c4bd.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.","maintainedGuideLinks":[],"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":[{"id":"term-llm","name":"Large language model","explanation":"A language model with many learned numerical settings, used to process and generate language.","referenceUrl":"https://developers.google.com/machine-learning/glossary#large-language-model"}],"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"},{"id":"time","name":"TIME","kind":"organization","context":"TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.","referenceUrls":["https://time.com/"],"path":"/entities/time","json":"/entities/time.json","markdown":"/entities/time.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"94e3c4bddc89872e084bb3a9b03a506e31724fa2ec69c0318d38ea0ee2564835","revision":"1f14ffde-964c-4397-8f12-94e7b839eb02","revisionKind":"publication-id","title":"Antonio Collante and coauthors on the impact of LLMs on code review","url":"https://escapemeridian.com/readings/antonio-collante-and-coauthors-on-the-impact-of-llms-on-code-94e3c4bd"},"source":{"type":"source","id":"https://arxiv.org/abs/2508.11034","revision":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","revisionKind":"sha256-source-text","title":"The Impact of Large Language Models (LLMs) on Code Review Process","url":"https://arxiv.org/abs/2508.11034"},"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":"94e3c4bddc89872e084bb3a9b03a506e31724fa2ec69c0318d38ea0ee2564835","title":"Antonio Collante and coauthors on the impact of LLMs on code review","argument":"Antonio Collante and coauthors examine GPT-assisted GitHub pull request (PR) workflows, asking whether assistance is associated with faster code review and shorter waits for acceptance. They analyzed 25,473 PRs from 9,254 GitHub projects, identifying GPT-assisted PRs with a semi-automated method they say reached 95% labeling accuracy. The authors report that GPT-assisted PRs had a median resolution time of 9 hours, compared with 23 hours for non-assisted PRs. They also report 33% less review time and an 87% reduction in waiting time before acceptance. In a separate sample of 300 GPT-assisted PRs, they say developers mainly used GPT for code optimization, bug fixing and documentation updates. The paper’s practical point is that GPT assistance may affect different stages of review, not just the act of reviewing code. The findings give software teams concrete workflow measures to examine. Escape Meridian interpretation: the abstract describes a dataset comparison and a heuristic for identifying assisted PRs; it does not establish that GPT alone caused the reported differences.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The abstract reports a semi-automated identification method with 95% labeling accuracy; it does not describe in detail how misidentified PRs might affect the comparisons. The abstract does not establish that GPT alone caused the reported time differences.","speaker":{"name":"Collante, Antonio, Abedu, Samuel, Khatoonabadi, SayedHassan, Abdellatif, Ahmad, Alor, Ebube, Shihab, Emad","roleAtSource":"Article author"},"source":{"title":"The Impact of Large Language Models (LLMs) on Code Review Process","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2508.11034","publishedAt":"2025-08-14","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors analyzed 25,473 pull requests from 9,254 GitHub projects and used a semi-automated heuristic to identify GPT-assisted pull requests, reporting 95% labeling accuracy.","excerpt":"We curated a dataset of 25,473 PRs from 9,254 GitHub projects and identified GPT-assisted PRs using a semi-automated heuristic approach that combines keyword-based detection, regular expression filtering, and manual verification until achieving 95% labeling accuracy.","url":"https://arxiv.org/abs/2508.11034","relationship":"Source-stated"},{"text":"The authors report median resolution time of 9 hours for GPT-assisted PRs and 23 hours for non-assisted PRs, a reduction of more than 60%.","excerpt":"GPT-assisted PRs reduced median resolution time by more than 60% (9 hours compared to 23 hours for non-assisted PRs).","url":"https://arxiv.org/abs/2508.11034","relationship":"Source-stated"},{"text":"The authors report that GPT use was associated with 33% less review time and 87% less waiting time before acceptance.","excerpt":"We discovered that utilizing GPT can reduce the review time by 33% and the waiting time before acceptance by 87%.","url":"https://arxiv.org/abs/2508.11034","relationship":"Source-stated"},{"text":"In a sample of 300 GPT-assisted PRs, the authors say developers mainly used GPT for code optimization, bug fixing and documentation updates.","excerpt":"developers predominantly use GPT for code optimization (60%), bug fixing (26%), and documentation updates (12%).","url":"https://arxiv.org/abs/2508.11034","relationship":"Source-stated"},{"text":"Because the abstract describes comparisons between identified GPT-assisted and non-assisted PRs, it does not establish that GPT alone caused the reported time differences.","excerpt":"to evaluate differences between GPT-assisted and non-assisted PRs, both at the overall resolution level and across distinct review phases.","url":"https://arxiv.org/abs/2508.11034","relationship":"Escape Meridian interpretation"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2508.11034","checkedAt":"2026-11-13T10:20:08.246Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Would the reported time differences hold if GPT-assisted PRs were identified or compared using a different method?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Antonio Collante and coauthors on the impact of LLMs on code review","summary":"Antonio Collante and coauthors examine GPT-assisted GitHub pull request (PR) workflows, asking whether assistance is associated with faster code review and shorter waits for acceptance. They analyzed 25,473 PRs from 9,254 GitHub projects, identifying GPT-assisted PRs with a semi-automated method they say reached 95% labeling accuracy. The authors report that GPT-assisted PRs had a median resolution time of 9 hours, compared with 23 hours for non-assisted PRs. They also report 33% less review time and an 87% reduction in waiting time before acceptance. In a separate sample of 300 GPT-assisted PRs, they say developers mainly used GPT for code optimization, bug fixing and documentation updates. The paper’s practical point is that GPT assistance may affect different stages of review, not just the act of reviewing code. The findings give software teams concrete workflow measures to examine. Escape Meridian interpretation: the abstract describes a dataset comparison and a heuristic for identifying assisted PRs; it does not establish that GPT alone caused the reported differences.","question":"Would the reported time differences hold if GPT-assisted PRs were identified or compared using a different method?","claims":[{"text":"The authors analyzed 25,473 pull requests from 9,254 GitHub projects and used a semi-automated heuristic to identify GPT-assisted pull requests, reporting 95% labeling accuracy.","excerpt":"We curated a dataset of 25,473 PRs from 9,254 GitHub projects and identified GPT-assisted PRs using a semi-automated heuristic approach that combines keyword-based detection, regular expression filtering, and manual verification until achieving 95% labeling accuracy.","support":"source-stated"},{"text":"The authors report median resolution time of 9 hours for GPT-assisted PRs and 23 hours for non-assisted PRs, a reduction of more than 60%.","excerpt":"GPT-assisted PRs reduced median resolution time by more than 60% (9 hours compared to 23 hours for non-assisted PRs).","support":"source-stated"},{"text":"The authors report that GPT use was associated with 33% less review time and 87% less waiting time before acceptance.","excerpt":"We discovered that utilizing GPT can reduce the review time by 33% and the waiting time before acceptance by 87%.","support":"source-stated"},{"text":"In a sample of 300 GPT-assisted PRs, the authors say developers mainly used GPT for code optimization, bug fixing and documentation updates.","excerpt":"developers predominantly use GPT for code optimization (60%), bug fixing (26%), and documentation updates (12%).","support":"source-stated"},{"text":"Because the abstract describes comparisons between identified GPT-assisted and non-assisted PRs, it does not establish that GPT alone caused the reported time differences.","excerpt":"to evaluate differences between GPT-assisted and non-assisted PRs, both at the overall resolution level and across distinct review phases.","support":"interpretation"}],"limitations":["The abstract reports a semi-automated identification method with 95% labeling accuracy; it does not describe in detail how misidentified PRs might affect the comparisons.","The abstract does not establish that GPT alone caused the reported time differences."],"mediaId":"6bfd02fd08ab63389c774e86325df7d1c0b8e643893839a69ecd112f844434f2","correctionReason":null,"id":"94e3c4bddc89872e084bb3a9b03a506e31724fa2ec69c0318d38ea0ee2564835","publicationId":"1f14ffde-964c-4397-8f12-94e7b839eb02","version":1,"sourceUrl":"https://arxiv.org/abs/2508.11034","sourceTitle":"The Impact of Large Language Models (LLMs) on Code Review Process","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2025-08-14","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-13T07:43:23.424Z","createdAt":"2026-11-13T09:16:46.707Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"6bfd02fd08ab63389c774e86325df7d1c0b8e643893839a69ecd112f844434f2","kind":"article","url":"https://arxiv.org/abs/2508.11034","sourceUrl":"https://arxiv.org/abs/2508.11034","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-13T09:04:04.766Z","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/2508.11034","checkedAt":"2026-11-13T10:20:08.246Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v6","mediaId":"6bfd02fd08ab63389c774e86325df7d1c0b8e643893839a69ecd112f844434f2","confidence":0.93,"provider":"typesafe","model":"jev-1.13.0","inputHash":"ef03eedb85cf44d8635f5865a4d9b4210cb716e38a31a4da10c76e2ef4993940","decidedAt":"2026-11-13T09:04:07.440Z"},"provenance":{"textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observationId":"c457ff23a7a9dbb84e68937d89dadad71ed8e38ff1077a76cd07a64d25c1bee0","retrievedAt":"2026-11-13T09:04:04.465Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"db3a8883-b34d-455e-849c-48e167bee9ab"},"reviewer":{"model":"gpt-6-luna","runId":"2320aeca-a17a-44f9-8372-86ddade3c845"}},"audit":{"status":"passed","reviewedAt":"2026-11-13T09:16:46.707Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Collante, Antonio","url":null,"kind":"unknown"},{"name":"Abedu, Samuel","url":null,"kind":"unknown"},{"name":"Khatoonabadi, SayedHassan","url":null,"kind":"unknown"},{"name":"Abdellatif, Ahmad","url":null,"kind":"unknown"},{"name":"Alor, Ebube","url":null,"kind":"unknown"},{"name":"Shihab, Emad","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Collante, Antonio","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Collante, Antonio","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Abedu, Samuel","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Abedu, Samuel","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Khatoonabadi, SayedHassan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Khatoonabadi, SayedHassan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Abdellatif, Ahmad","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Abdellatif, Ahmad","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Alor, Ebube","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Alor, Ebube","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}},{"actor":{"name":"Shihab, Emad","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Shihab, Emad","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2508.11034","locator":"head/meta[name=author|citation_author]"},"assertionId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T10:20:08.246Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","sourceUrl":"https://arxiv.org/abs/2508.11034","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z"}}],"participationCoverage":{"total":6,"returned":6,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2508.11034","snapshotId":"81f5c4d3629df73b5e5951266fc84440638cbf088118943fe4d47e6c767a661d","contentSha256":"25e6c167f18ef16c123d4631634373d2b6e292328967453b4ebd297f2d8b023d","textHash":"b2a09a2c8dd1399d0f8a87473012fbf2779f032e64cbb96851870666cc0be57d","observedAt":"2026-11-13T10:20:08.246Z","basis":"publisher-page-metadata","sourceLinks":48,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/94e3c4bddc89872e084bb3a9b03a506e31724fa2ec69c0318d38ea0ee2564835/source"}}}],"audits":[{"publicationId":"1f14ffde-964c-4397-8f12-94e7b839eb02","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading attributes the study to Antonio Collante and coauthors, matching the listed authors. Findings are framed as what the authors report or say, so they are not presented as established fact.","excerpt":"Authors:Antonio Collante, Samuel Abedu, SayedHassan Khatoonabadi, Ahmad Abdellatif, Ebube Alor, Emad Shihab"},{"dimension":"summary","status":"pass","explanation":"The title, summary and question faithfully convey the study's aim, dataset, method and headline numbers. The PR abbreviation is defined, and the phase-specific framing reflects the abstract. The causation caveat is labeled as Escape Meridian interpretation. Minor gap: the authors' stronger claim that early GPT adoption can substantially boost PR effectiveness is softened, but it is not erased in a misleading way. Terms such as pull request and median resolution time are suited to tooltips.","excerpt":"the phase-specific effects of LLM assistance on the efficiency of code review processes remain underexplored"},{"dimension":"date","status":"pass","explanation":"originalPublished is 2025-08-14, which matches the v1 submission date. The later v2 revision in November 2025 and the 2026 feed date do not change the original publication date.","excerpt":"[Submitted on 14 Aug 2025 (v1), last revised 13 Nov 2025 (this version, v2)]"},{"dimension":"support","status":"pass","explanation":"Every numeric claim (25,473 PRs, 9,254 projects, 95% accuracy, 9 vs 23 hours, 33%, 87%, 300-PR sample categories) matches the abstract verbatim. The non-causation point is labeled as interpretation and grounded in the abstract's description of comparative statistical methods. The note on actionable workflow measures echoes the authors' own framing. No unsupported benefits or costs are added.","excerpt":"We then applied statistical modeling, including multiple linear regression and Mann-Whitney U test, to evaluate differences between GPT-assisted and non-assisted PRs"},{"dimension":"media","status":"pass","explanation":"The selected media is the verified arXiv abstract page for this paper. It is the direct original source and the most relevant link.","excerpt":null}],"createdAt":"2026-11-13T09:16:46.707Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}