{"id":"c2bc7a6492cf486b4dd4942780bd327b9b1ae2f482f153353925239fae36b8ec","title":"Brynjolfsson, Li and Raymond study generative AI at work","argument":"In “Generative AI at Work,” Erik Brynjolfsson, Danielle Li and Lindsey R. Raymond study a generative AI-based conversational assistant introduced in stages to customer support agents. Using data from 5,179 agents, they report that access to the tool increased issues resolved per hour by 14% on average. The reported gain was 34% for novice and low-skilled workers, while experienced and highly skilled workers saw minimal impact.\n\nThe authors offer suggestive evidence that the assistant spreads best practices from more able workers and helps newer workers move down the experience curve. They also report improved customer sentiment and employee retention, and say the tool may support worker learning. Their argument is not that every worker benefits equally: the measured productivity effects vary substantially by workers’ experience and skill. The results make the distribution of gains central to understanding AI assistance at work, alongside its average effect. The evidence described on the NBER page comes from a particular customer-support setting; it does not establish whether similar effects hold in other jobs or workplaces.\n\nHow much of the reported difference between novice and experienced workers reflects the assistant spreading best practices, rather than other differences in how workers use it?","topicIds":["work-and-agency","power-and-infrastructure"],"domain":"work","claimType":"attributed reading","limits":"The evidence described on the NBER page concerns 5,179 customer support agents using a conversational assistant; the page does not establish whether similar effects hold in other jobs or workplaces. The proposed explanation that the assistant spreads best practices is described by the authors as suggestive evidence, not a demonstrated mechanism.","speaker":{"name":"Erik Brynjolfsson, Danielle Li, Lindsey R. Raymond","roleAtSource":"Article author"},"source":{"title":"Generative AI at Work","publisher":"National Bureau of Economic Research — new working papers","format":"Article","url":"https://www.nber.org/papers/w31161","publishedAt":"2023-04","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors report a 14% average increase in issues resolved per hour after agents gained access to the conversational assistant.","excerpt":"Access to the tool increases productivity, as measured by issues resolved per hour, by 14% on average","url":"https://www.nber.org/papers/w31161","relationship":"Source-stated"},{"text":"The reported productivity effect was larger for novice and low-skilled workers than for experienced and highly skilled workers.","excerpt":"including a 34% improvement for novice and low-skilled workers but with minimal impact on experienced and highly skilled workers","url":"https://www.nber.org/papers/w31161","relationship":"Source-stated"},{"text":"The authors offer suggestive evidence that the AI model spreads best practices and helps newer workers gain experience.","excerpt":"We provide suggestive evidence that the AI model disseminates the best practices of more able workers and helps newer workers move down the experience curve.","url":"https://www.nber.org/papers/w31161","relationship":"Source-stated"},{"text":"The authors report improved customer sentiment and employee retention, and say the tool may support worker learning.","excerpt":"AI assistance improves customer sentiment, increases employee retention, and may lead to worker learning.","url":"https://www.nber.org/papers/w31161","relationship":"Source-stated"}],"sourceMedia":{"status":"found","sourceUrl":"https://www.nber.org/papers/w31161","checkedAt":"2026-11-05T15:54:15.532Z","items":[{"type":"image","url":"https://www.nber.org/sites/default/files/2022-06/NBER-FB-Share-Tile-1200.jpg","sourceUrl":"https://www.nber.org/papers/w31161","attribution":"National Bureau of Economic Research — new working papers","basis":"publisher-metadata","availability":"verified","verifiedAt":"2026-11-05T15:54:15.875Z"}],"note":"Media declared by the original publisher; availability checked separately from the editorial reading."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How much of the reported difference between novice and experienced workers reflects the assistant spreading best practices, rather than other differences in how workers use it?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Brynjolfsson, Li and Raymond study generative AI at work","summary":"In “Generative AI at Work,” Erik Brynjolfsson, Danielle Li and Lindsey R. Raymond study a generative AI-based conversational assistant introduced in stages to customer support agents. Using data from 5,179 agents, they report that access to the tool increased issues resolved per hour by 14% on average. The reported gain was 34% for novice and low-skilled workers, while experienced and highly skilled workers saw minimal impact.\n\nThe authors offer suggestive evidence that the assistant spreads best practices from more able workers and helps newer workers move down the experience curve. They also report improved customer sentiment and employee retention, and say the tool may support worker learning. Their argument is not that every worker benefits equally: the measured productivity effects vary substantially by workers’ experience and skill. The results make the distribution of gains central to understanding AI assistance at work, alongside its average effect. The evidence described on the NBER page comes from a particular customer-support setting; it does not establish whether similar effects hold in other jobs or workplaces.\n\nHow much of the reported difference between novice and experienced workers reflects the assistant spreading best practices, rather than other differences in how workers use it?","question":"How much of the reported difference between novice and experienced workers reflects the assistant spreading best practices, rather than other differences in how workers use it?","claims":[{"text":"The authors report a 14% average increase in issues resolved per hour after agents gained access to the conversational assistant.","excerpt":"Access to the tool increases productivity, as measured by issues resolved per hour, by 14% on average","support":"source-stated"},{"text":"The reported productivity effect was larger for novice and low-skilled workers than for experienced and highly skilled workers.","excerpt":"including a 34% improvement for novice and low-skilled workers but with minimal impact on experienced and highly skilled workers","support":"source-stated"},{"text":"The authors offer suggestive evidence that the AI model spreads best practices and helps newer workers gain experience.","excerpt":"We provide suggestive evidence that the AI model disseminates the best practices of more able workers and helps newer workers move down the experience curve.","support":"source-stated"},{"text":"The authors report improved customer sentiment and employee retention, and say the tool may support worker learning.","excerpt":"AI assistance improves customer sentiment, increases employee retention, and may lead to worker learning.","support":"source-stated"}],"limitations":["The evidence described on the NBER page concerns 5,179 customer support agents using a conversational assistant; the page does not establish whether similar effects hold in other jobs or workplaces.","The proposed explanation that the assistant spreads best practices is described by the authors as suggestive evidence, not a demonstrated mechanism."],"mediaId":"92735c509f6627dd7ca95456196f2934a91874fbe0f3736a32a44f25b5a2d45c","correctionReason":null,"id":"c2bc7a6492cf486b4dd4942780bd327b9b1ae2f482f153353925239fae36b8ec","publicationId":"3fb17aa5-5b84-443e-8fd1-084f56c91eac","version":1,"sourceUrl":"https://www.nber.org/papers/w31161","sourceTitle":"Generative AI at Work","publisher":"National Bureau of Economic Research — new working papers","publishedAt":null,"originalPublished":{"value":"2023-04","precision":"month","basis":"stated_with_named_work"},"firstObservedAt":"2026-11-05T15:41:56.887Z","createdAt":"2026-11-05T15:55:09.623Z","historical":true,"entityIds":["nber"],"topicIds":["work-and-agency","power-and-infrastructure"],"media":[{"id":"92735c509f6627dd7ca95456196f2934a91874fbe0f3736a32a44f25b5a2d45c","kind":"article","url":"https://www.nber.org/papers/w31161","sourceUrl":"https://www.nber.org/papers/w31161","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-05T15:54:15.875Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."},{"id":"8bd07506376f662ba07acb4fd2718ac939cd082448f783faf89590009122c55a","kind":"image","url":"https://www.nber.org/sites/default/files/2022-06/NBER-FB-Share-Tile-1200.jpg","sourceUrl":"https://www.nber.org/papers/w31161","label":"View source image","provenance":"publisher_page","verification":{"status":"verified","checkedAt":"2026-11-05T15:54:15.875Z","method":"publisher_association_and_head","contentType":"image/jpeg","cacheUntil":1793894355875},"recommendationReason":"Observed on the acquired publisher source; availability checked."}],"sourceMedia":{"status":"found","sourceUrl":"https://www.nber.org/papers/w31161","checkedAt":"2026-11-05T15:54:15.532Z","items":[{"type":"image","url":"https://www.nber.org/sites/default/files/2022-06/NBER-FB-Share-Tile-1200.jpg","sourceUrl":"https://www.nber.org/papers/w31161","attribution":"National Bureau of Economic Research — new working papers","basis":"publisher-metadata","availability":"verified","verifiedAt":"2026-11-05T15:54:15.875Z"}],"note":"Media declared by the original publisher; availability checked separately from the editorial reading."},"editorialDecision":{"outcome":"accept","reasonCodes":["useful_perspective"],"rubricVersion":"stream-admission-v5","mediaId":"92735c509f6627dd7ca95456196f2934a91874fbe0f3736a32a44f25b5a2d45c","confidence":1,"provider":"typesafe","model":"jev-1.13.0","inputHash":"ea957dee794bca184db7f2dde0b4323a6390570e41bd34ff24cf5e00228d5cfe","decidedAt":"2026-11-05T15:54:18.186Z"},"provenance":{"textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observationId":"e8edabeff2a79ec98c484786c76e536aed7ebfba9373d91405c99836ffe434d5","retrievedAt":"2026-11-05T15:54:15.532Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"0569721d-c3b1-4c4a-8168-a829b5d80d1a"},"reviewer":{"model":"gpt-6-luna","runId":"6369f828-107c-4495-81e9-d25a62048eee"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-05T15:55:09.623Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Erik Brynjolfsson","url":null,"kind":"unknown"},{"name":"Danielle Li","url":null,"kind":"unknown"},{"name":"Lindsey R. Raymond","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Erik Brynjolfsson","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Erik Brynjolfsson","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://www.nber.org/papers/w31161","locator":"head/meta[name=author|citation_author]"},"assertionId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T15:54:15.532Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4","sourceUrl":"https://www.nber.org/papers/w31161","contentSha256":"1461eeb678c9a133a2ba140b7c53157d39d9fe6d2ffcdb47fa06dfbece0609f2","textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observedAt":"2026-11-05T15:54:15.532Z"}},{"actor":{"name":"Danielle Li","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Danielle Li","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://www.nber.org/papers/w31161","locator":"head/meta[name=author|citation_author]"},"assertionId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T15:54:15.532Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4","sourceUrl":"https://www.nber.org/papers/w31161","contentSha256":"1461eeb678c9a133a2ba140b7c53157d39d9fe6d2ffcdb47fa06dfbece0609f2","textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observedAt":"2026-11-05T15:54:15.532Z"}},{"actor":{"name":"Lindsey R. Raymond","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lindsey R. Raymond","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://www.nber.org/papers/w31161","locator":"head/meta[name=author|citation_author]"},"assertionId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T15:54:15.532Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4","sourceUrl":"https://www.nber.org/papers/w31161","contentSha256":"1461eeb678c9a133a2ba140b7c53157d39d9fe6d2ffcdb47fa06dfbece0609f2","textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observedAt":"2026-11-05T15:54:15.532Z"}}],"participationCoverage":{"total":3,"returned":3,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://www.nber.org/papers/w31161","snapshotId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4","contentSha256":"1461eeb678c9a133a2ba140b7c53157d39d9fe6d2ffcdb47fa06dfbece0609f2","textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observedAt":"2026-11-05T15:54:15.532Z","basis":"publisher-page-metadata","sourceLinks":49,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/c2bc7a6492cf486b4dd4942780bd327b9b1ae2f482f153353925239fae36b8ec/source"}},"canonicalUrl":"https://escapemeridian.com/readings/brynjolfsson-li-and-raymond-study-generative-ai-at-work-c2bc7a64","formats":{"html":"/readings/brynjolfsson-li-and-raymond-study-generative-ai-at-work-c2bc7a64","markdown":"/readings/brynjolfsson-li-and-raymond-study-generative-ai-at-work-c2bc7a64.md","json":"/readings/brynjolfsson-li-and-raymond-study-generative-ai-at-work-c2bc7a64.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":[],"entities":[{"id":"nber","name":"National Bureau of Economic Research","kind":"organization","context":"The National Bureau of Economic Research shares economic research, including working papers on technology and work.","referenceUrls":["https://www.nber.org/"],"path":"/entities/nber","json":"/entities/nber.json","markdown":"/entities/nber.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"c2bc7a6492cf486b4dd4942780bd327b9b1ae2f482f153353925239fae36b8ec","revision":"3fb17aa5-5b84-443e-8fd1-084f56c91eac","revisionKind":"publication-id","title":"Brynjolfsson, Li and Raymond study generative AI at work","url":"https://escapemeridian.com/readings/brynjolfsson-li-and-raymond-study-generative-ai-at-work-c2bc7a64"},"source":{"type":"source","id":"https://www.nber.org/papers/w31161","revision":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","revisionKind":"sha256-source-text","title":"Generative AI at Work","url":"https://www.nber.org/papers/w31161"},"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":"c2bc7a6492cf486b4dd4942780bd327b9b1ae2f482f153353925239fae36b8ec","title":"Brynjolfsson, Li and Raymond study generative AI at work","argument":"In “Generative AI at Work,” Erik Brynjolfsson, Danielle Li and Lindsey R. Raymond study a generative AI-based conversational assistant introduced in stages to customer support agents. Using data from 5,179 agents, they report that access to the tool increased issues resolved per hour by 14% on average. The reported gain was 34% for novice and low-skilled workers, while experienced and highly skilled workers saw minimal impact.\n\nThe authors offer suggestive evidence that the assistant spreads best practices from more able workers and helps newer workers move down the experience curve. They also report improved customer sentiment and employee retention, and say the tool may support worker learning. Their argument is not that every worker benefits equally: the measured productivity effects vary substantially by workers’ experience and skill. The results make the distribution of gains central to understanding AI assistance at work, alongside its average effect. The evidence described on the NBER page comes from a particular customer-support setting; it does not establish whether similar effects hold in other jobs or workplaces.\n\nHow much of the reported difference between novice and experienced workers reflects the assistant spreading best practices, rather than other differences in how workers use it?","topicIds":["work-and-agency","power-and-infrastructure"],"domain":"work","claimType":"attributed reading","limits":"The evidence described on the NBER page concerns 5,179 customer support agents using a conversational assistant; the page does not establish whether similar effects hold in other jobs or workplaces. The proposed explanation that the assistant spreads best practices is described by the authors as suggestive evidence, not a demonstrated mechanism.","speaker":{"name":"Erik Brynjolfsson, Danielle Li, Lindsey R. Raymond","roleAtSource":"Article author"},"source":{"title":"Generative AI at Work","publisher":"National Bureau of Economic Research — new working papers","format":"Article","url":"https://www.nber.org/papers/w31161","publishedAt":"2023-04","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors report a 14% average increase in issues resolved per hour after agents gained access to the conversational assistant.","excerpt":"Access to the tool increases productivity, as measured by issues resolved per hour, by 14% on average","url":"https://www.nber.org/papers/w31161","relationship":"Source-stated"},{"text":"The reported productivity effect was larger for novice and low-skilled workers than for experienced and highly skilled workers.","excerpt":"including a 34% improvement for novice and low-skilled workers but with minimal impact on experienced and highly skilled workers","url":"https://www.nber.org/papers/w31161","relationship":"Source-stated"},{"text":"The authors offer suggestive evidence that the AI model spreads best practices and helps newer workers gain experience.","excerpt":"We provide suggestive evidence that the AI model disseminates the best practices of more able workers and helps newer workers move down the experience curve.","url":"https://www.nber.org/papers/w31161","relationship":"Source-stated"},{"text":"The authors report improved customer sentiment and employee retention, and say the tool may support worker learning.","excerpt":"AI assistance improves customer sentiment, increases employee retention, and may lead to worker learning.","url":"https://www.nber.org/papers/w31161","relationship":"Source-stated"}],"sourceMedia":{"status":"found","sourceUrl":"https://www.nber.org/papers/w31161","checkedAt":"2026-11-05T15:54:15.532Z","items":[{"type":"image","url":"https://www.nber.org/sites/default/files/2022-06/NBER-FB-Share-Tile-1200.jpg","sourceUrl":"https://www.nber.org/papers/w31161","attribution":"National Bureau of Economic Research — new working papers","basis":"publisher-metadata","availability":"verified","verifiedAt":"2026-11-05T15:54:15.875Z"}],"note":"Media declared by the original publisher; availability checked separately from the editorial reading."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How much of the reported difference between novice and experienced workers reflects the assistant spreading best practices, rather than other differences in how workers use it?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Brynjolfsson, Li and Raymond study generative AI at work","summary":"In “Generative AI at Work,” Erik Brynjolfsson, Danielle Li and Lindsey R. Raymond study a generative AI-based conversational assistant introduced in stages to customer support agents. Using data from 5,179 agents, they report that access to the tool increased issues resolved per hour by 14% on average. The reported gain was 34% for novice and low-skilled workers, while experienced and highly skilled workers saw minimal impact.\n\nThe authors offer suggestive evidence that the assistant spreads best practices from more able workers and helps newer workers move down the experience curve. They also report improved customer sentiment and employee retention, and say the tool may support worker learning. Their argument is not that every worker benefits equally: the measured productivity effects vary substantially by workers’ experience and skill. The results make the distribution of gains central to understanding AI assistance at work, alongside its average effect. The evidence described on the NBER page comes from a particular customer-support setting; it does not establish whether similar effects hold in other jobs or workplaces.\n\nHow much of the reported difference between novice and experienced workers reflects the assistant spreading best practices, rather than other differences in how workers use it?","question":"How much of the reported difference between novice and experienced workers reflects the assistant spreading best practices, rather than other differences in how workers use it?","claims":[{"text":"The authors report a 14% average increase in issues resolved per hour after agents gained access to the conversational assistant.","excerpt":"Access to the tool increases productivity, as measured by issues resolved per hour, by 14% on average","support":"source-stated"},{"text":"The reported productivity effect was larger for novice and low-skilled workers than for experienced and highly skilled workers.","excerpt":"including a 34% improvement for novice and low-skilled workers but with minimal impact on experienced and highly skilled workers","support":"source-stated"},{"text":"The authors offer suggestive evidence that the AI model spreads best practices and helps newer workers gain experience.","excerpt":"We provide suggestive evidence that the AI model disseminates the best practices of more able workers and helps newer workers move down the experience curve.","support":"source-stated"},{"text":"The authors report improved customer sentiment and employee retention, and say the tool may support worker learning.","excerpt":"AI assistance improves customer sentiment, increases employee retention, and may lead to worker learning.","support":"source-stated"}],"limitations":["The evidence described on the NBER page concerns 5,179 customer support agents using a conversational assistant; the page does not establish whether similar effects hold in other jobs or workplaces.","The proposed explanation that the assistant spreads best practices is described by the authors as suggestive evidence, not a demonstrated mechanism."],"mediaId":"92735c509f6627dd7ca95456196f2934a91874fbe0f3736a32a44f25b5a2d45c","correctionReason":null,"id":"c2bc7a6492cf486b4dd4942780bd327b9b1ae2f482f153353925239fae36b8ec","publicationId":"3fb17aa5-5b84-443e-8fd1-084f56c91eac","version":1,"sourceUrl":"https://www.nber.org/papers/w31161","sourceTitle":"Generative AI at Work","publisher":"National Bureau of Economic Research — new working papers","publishedAt":null,"originalPublished":{"value":"2023-04","precision":"month","basis":"stated_with_named_work"},"firstObservedAt":"2026-11-05T15:41:56.887Z","createdAt":"2026-11-05T15:55:09.623Z","historical":true,"entityIds":["nber"],"topicIds":["work-and-agency","power-and-infrastructure"],"media":[{"id":"92735c509f6627dd7ca95456196f2934a91874fbe0f3736a32a44f25b5a2d45c","kind":"article","url":"https://www.nber.org/papers/w31161","sourceUrl":"https://www.nber.org/papers/w31161","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-05T15:54:15.875Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."},{"id":"8bd07506376f662ba07acb4fd2718ac939cd082448f783faf89590009122c55a","kind":"image","url":"https://www.nber.org/sites/default/files/2022-06/NBER-FB-Share-Tile-1200.jpg","sourceUrl":"https://www.nber.org/papers/w31161","label":"View source image","provenance":"publisher_page","verification":{"status":"verified","checkedAt":"2026-11-05T15:54:15.875Z","method":"publisher_association_and_head","contentType":"image/jpeg","cacheUntil":1793894355875},"recommendationReason":"Observed on the acquired publisher source; availability checked."}],"sourceMedia":{"status":"found","sourceUrl":"https://www.nber.org/papers/w31161","checkedAt":"2026-11-05T15:54:15.532Z","items":[{"type":"image","url":"https://www.nber.org/sites/default/files/2022-06/NBER-FB-Share-Tile-1200.jpg","sourceUrl":"https://www.nber.org/papers/w31161","attribution":"National Bureau of Economic Research — new working papers","basis":"publisher-metadata","availability":"verified","verifiedAt":"2026-11-05T15:54:15.875Z"}],"note":"Media declared by the original publisher; availability checked separately from the editorial reading."},"editorialDecision":{"outcome":"accept","reasonCodes":["useful_perspective"],"rubricVersion":"stream-admission-v5","mediaId":"92735c509f6627dd7ca95456196f2934a91874fbe0f3736a32a44f25b5a2d45c","confidence":1,"provider":"typesafe","model":"jev-1.13.0","inputHash":"ea957dee794bca184db7f2dde0b4323a6390570e41bd34ff24cf5e00228d5cfe","decidedAt":"2026-11-05T15:54:18.186Z"},"provenance":{"textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observationId":"e8edabeff2a79ec98c484786c76e536aed7ebfba9373d91405c99836ffe434d5","retrievedAt":"2026-11-05T15:54:15.532Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"0569721d-c3b1-4c4a-8168-a829b5d80d1a"},"reviewer":{"model":"gpt-6-luna","runId":"6369f828-107c-4495-81e9-d25a62048eee"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-05T15:55:09.623Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Erik Brynjolfsson","url":null,"kind":"unknown"},{"name":"Danielle Li","url":null,"kind":"unknown"},{"name":"Lindsey R. Raymond","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Erik Brynjolfsson","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Erik Brynjolfsson","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://www.nber.org/papers/w31161","locator":"head/meta[name=author|citation_author]"},"assertionId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T15:54:15.532Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4","sourceUrl":"https://www.nber.org/papers/w31161","contentSha256":"1461eeb678c9a133a2ba140b7c53157d39d9fe6d2ffcdb47fa06dfbece0609f2","textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observedAt":"2026-11-05T15:54:15.532Z"}},{"actor":{"name":"Danielle Li","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Danielle Li","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://www.nber.org/papers/w31161","locator":"head/meta[name=author|citation_author]"},"assertionId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T15:54:15.532Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4","sourceUrl":"https://www.nber.org/papers/w31161","contentSha256":"1461eeb678c9a133a2ba140b7c53157d39d9fe6d2ffcdb47fa06dfbece0609f2","textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observedAt":"2026-11-05T15:54:15.532Z"}},{"actor":{"name":"Lindsey R. Raymond","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lindsey R. Raymond","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://www.nber.org/papers/w31161","locator":"head/meta[name=author|citation_author]"},"assertionId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T15:54:15.532Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4","sourceUrl":"https://www.nber.org/papers/w31161","contentSha256":"1461eeb678c9a133a2ba140b7c53157d39d9fe6d2ffcdb47fa06dfbece0609f2","textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observedAt":"2026-11-05T15:54:15.532Z"}}],"participationCoverage":{"total":3,"returned":3,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://www.nber.org/papers/w31161","snapshotId":"4ed4b5c650dab996bb953e4f029823ce16f699e381e489419e738bd3d0ac22a4","contentSha256":"1461eeb678c9a133a2ba140b7c53157d39d9fe6d2ffcdb47fa06dfbece0609f2","textHash":"05759a470cc48187fe1302a61c28892eea4d56a3b81e01543b9d02baf8671f38","observedAt":"2026-11-05T15:54:15.532Z","basis":"publisher-page-metadata","sourceLinks":49,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/c2bc7a6492cf486b4dd4942780bd327b9b1ae2f482f153353925239fae36b8ec/source"}}}],"audits":[{"publicationId":"3fb17aa5-5b84-443e-8fd1-084f56c91eac","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading correctly attributes the paper and its findings to Erik Brynjolfsson, Danielle Li and Lindsey R. Raymond. Throughout, it presents the results as the authors' reported findings rather than established fact.","excerpt":"Erik Brynjolfsson, Danielle Li, and Lindsey R. Raymond, \"Generative AI at Work,\" NBER Working Paper 31161 (2023)"},{"dimension":"summary","status":"pass","explanation":"The title, summary and question faithfully convey the study setting, the 14% average gain, the 34% gain for novices with minimal effect for experienced workers, and the heterogeneity. They also convey the suggestive mechanism and the limits of scope. The question follows from the suggestive-evidence caveat. A small improvement would be a plain-language tooltip for 'experience curve', but the meaning remains understandable.","excerpt":"Our results suggest that access to generative AI can increase productivity, with large heterogeneity in effects across workers."},{"dimension":"date","status":"pass","explanation":"The displayed date of April 2023 matches the paper's original issue date. The November 2023 revision and the 2025 journal publication are correctly not used as the original date.","excerpt":"Issue Date April 2023"},{"dimension":"support","status":"uncertain","explanation":"Most claims are source-stated, and the limitations are fair. However, claim 3 paraphrases 'helps newer workers move down the experience curve' as 'helps newer workers gain experience.' Moving down the experience curve means reaching experienced-level performance faster; it is not the same as gaining experience. This paraphrases a technical term. Smallest repair: keep the exact source phrase, as the summary already does.","excerpt":"helps newer workers move down the experience curve"},{"dimension":"media","status":"pass","explanation":"The selected media item is the verified direct link to the NBER working paper page. That page is the original source and the easiest way for readers to access it.","excerpt":null}],"createdAt":"2026-11-05T15:55:09.623Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}