{"id":"89333371ed8a8bbc2619ab81611cd65c2701b02df31817bd0f51b8f4b41faeaf","title":"Meredith Ringel Morris and coauthors propose an AGI framework at Google DeepMind","argument":"The paper was published on July 21, 2024. Meredith Ringel Morris and coauthors propose “Levels of AGI,” a framework for classifying the capabilities and behavior of AGI models and their precursors. They organize capability by performance and generality, and discuss how current systems fit the framework. The authors say the levels could provide a common language for comparing models, assessing risks, and measuring progress toward AGI. They also describe challenging requirements for future benchmarks to quantify systems’ behavior and capabilities against the levels. The framework connects capability to deployment considerations: the authors discuss how autonomy and risk interact with the levels and emphasize carefully selecting Human-AI Interaction paradigms for responsible and safe deployment. The abstract does not specify which benchmarks would meet those requirements or how the framework classifies particular current systems.","topicIds":["artificial-general-intelligence"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on October 28, 2026; originally published July 21, 2024. The available evidence is the publication page's abstract; it does not provide the framework's detailed levels or examples of how current systems fit them.","speaker":{"name":"Google DeepMind","roleAtSource":"Publisher"},"source":{"title":"Levels of AGI for Operationalizing Progress on the Path to AGI","publisher":"Google DeepMind","format":"Article","url":"https://deepmind.google/research/publications/66938/","publishedAt":"2024-07-21T00:00:00+00:00","dateBasis":"Date stated on the publisher’s page."},"evidence":[{"text":"Meredith Ringel Morris and coauthors propose a framework for classifying AGI models and their precursors by performance, generality, and autonomy.","excerpt":"We propose a framework for classifying the capabilities and behavior of Artificial General Intelligence (AGI) models and their precursors. This framework introduces levels of AGI performance, generality, and autonomy","url":"https://deepmind.google/research/publications/66938/","relationship":"Source-stated"},{"text":"The authors say future benchmarks face challenging requirements to quantify model behavior and capabilities against the levels.","excerpt":"We discuss the challenging requirements for future benchmarks that quantify the behavior and capabilities of AGI models against these levels.","url":"https://deepmind.google/research/publications/66938/","relationship":"Source-stated"},{"text":"The authors connect the levels to deployment considerations, including autonomy and risk, and emphasize careful selection of Human-AI Interaction paradigms.","excerpt":"Finally, we discuss how these levels of AGI interact with deployment considerations such as autonomy and risk, and emphasize the importance of carefully selecting Human-AI Interaction paradigms for responsible and safe deployment of highly capable AI systems.","url":"https://deepmind.google/research/publications/66938/","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://deepmind.google/research/publications/66938/","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What benchmark designs could quantify model behavior and capabilities against the proposed levels?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Meredith Ringel Morris and coauthors propose an AGI framework at Google DeepMind","summary":"The paper was published on July 21, 2024. Meredith Ringel Morris and coauthors propose “Levels of AGI,” a framework for classifying the capabilities and behavior of AGI models and their precursors. They organize capability by performance and generality, and discuss how current systems fit the framework. The authors say the levels could provide a common language for comparing models, assessing risks, and measuring progress toward AGI. They also describe challenging requirements for future benchmarks to quantify systems’ behavior and capabilities against the levels. The framework connects capability to deployment considerations: the authors discuss how autonomy and risk interact with the levels and emphasize carefully selecting Human-AI Interaction paradigms for responsible and safe deployment. The abstract does not specify which benchmarks would meet those requirements or how the framework classifies particular current systems.","question":"What benchmark designs could quantify model behavior and capabilities against the proposed levels?","claims":[{"text":"Meredith Ringel Morris and coauthors propose a framework for classifying AGI models and their precursors by performance, generality, and autonomy.","excerpt":"We propose a framework for classifying the capabilities and behavior of Artificial General Intelligence (AGI) models and their precursors. This framework introduces levels of AGI performance, generality, and autonomy","support":"source-stated"},{"text":"The authors say future benchmarks face challenging requirements to quantify model behavior and capabilities against the levels.","excerpt":"We discuss the challenging requirements for future benchmarks that quantify the behavior and capabilities of AGI models against these levels.","support":"source-stated"},{"text":"The authors connect the levels to deployment considerations, including autonomy and risk, and emphasize careful selection of Human-AI Interaction paradigms.","excerpt":"Finally, we discuss how these levels of AGI interact with deployment considerations such as autonomy and risk, and emphasize the importance of carefully selecting Human-AI Interaction paradigms for responsible and safe deployment of highly capable AI systems.","support":"source-stated"}],"limitations":["Text as retrieved on October 28, 2026; originally published July 21, 2024.","The available evidence is the publication page's abstract; it does not provide the framework's detailed levels or examples of how current systems fit them."],"mediaId":null,"correctionReason":null,"id":"89333371ed8a8bbc2619ab81611cd65c2701b02df31817bd0f51b8f4b41faeaf","publicationId":"874fdcb1-69ec-c10f-ea5c-3f9283271205","version":1,"sourceUrl":"https://deepmind.google/research/publications/66938/","sourceTitle":"Levels of AGI for Operationalizing Progress on the Path to AGI","publisher":"Google DeepMind","publishedAt":"2024-07-21T00:00:00+00:00","originalPublished":{"value":"2024-07-21T00:00:00+00:00","precision":"instant","basis":"backfill:article_published"},"firstObservedAt":"2026-10-28T07:39:56.365Z","createdAt":"2026-11-17T17:24:00.554Z","historical":true,"entityIds":["google-deepmind"],"topicIds":["artificial-general-intelligence"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://deepmind.google/research/publications/66938/","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"07846da93df64867a88446e29290c0a55141c275e72730e9ab09fb26a7ad9327","rubricVersion":"unrecorded","provider":"typesafe","model":"jev-1.13.0","reasonCodes":["bridge_admitted_by_model"],"confidence":0.82,"decidedAt":"2026-10-28T07:39:55.136Z"},"provenance":{"textHash":"ba2312eb7d8e6bb2886439f945659c048a2443281e7d3b129335e8c5a4c254df","observationId":"7422ad8eae23d56068d13586376db943f797ba131e993c6feea8113474a47a45","retrievedAt":"2026-10-28T07:41:48.620Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"8a2268f7-ff56-25c1-3e41-724d8e4fd566"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"39d870e4-c862-c9f6-5d3c-cd027c11be14"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-17T17:19:26.459Z","model":"anthropic/claude-opus-5.5"}},"canonicalUrl":"https://escapemeridian.com/readings/meredith-ringel-morris-and-coauthors-propose-an-agi-framework-89333371","formats":{"html":"/readings/meredith-ringel-morris-and-coauthors-propose-an-agi-framework-89333371","markdown":"/readings/meredith-ringel-morris-and-coauthors-propose-an-agi-framework-89333371.md","json":"/readings/meredith-ringel-morris-and-coauthors-propose-an-agi-framework-89333371.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":[],"conditions":[],"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-agi","name":"Artificial general intelligence","explanation":"The idea of AI with broad problem-solving, creativity and adaptability across different tasks.","referenceUrl":"https://developers.google.com/machine-learning/glossary#artificial-general-intelligence"},{"id":"term-benchmark","name":"Benchmark","explanation":"A defined set of tests used to compare systems. Results depend on the tasks and measures selected.","referenceUrl":"https://arxiv.org/abs/2211.09110"}],"entities":[{"id":"google-deepmind","name":"Google DeepMind","kind":"organization","context":"Google DeepMind develops AI systems for general tasks and scientific research, including AlphaFold.","referenceUrls":["https://deepmind.google/blog/","https://deepmind.google/science/alphafold/"],"path":"/entities/google-deepmind","json":"/entities/google-deepmind.json","markdown":"/entities/google-deepmind.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"89333371ed8a8bbc2619ab81611cd65c2701b02df31817bd0f51b8f4b41faeaf","revision":"874fdcb1-69ec-c10f-ea5c-3f9283271205","revisionKind":"publication-id","title":"Meredith Ringel Morris and coauthors propose an AGI framework at Google DeepMind","url":"https://escapemeridian.com/readings/meredith-ringel-morris-and-coauthors-propose-an-agi-framework-89333371"},"source":{"type":"source","id":"https://deepmind.google/research/publications/66938/","revision":"ba2312eb7d8e6bb2886439f945659c048a2443281e7d3b129335e8c5a4c254df","revisionKind":"sha256-source-text","title":"Levels of AGI for Operationalizing Progress on the Path to AGI","url":"https://deepmind.google/research/publications/66938/"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings to this reading. Shared topics are not citations. Other record types need an explicit reference adapter."},"revisions":[{"id":"89333371ed8a8bbc2619ab81611cd65c2701b02df31817bd0f51b8f4b41faeaf","title":"Meredith Ringel Morris and coauthors propose an AGI framework at Google DeepMind","argument":"The paper was published on July 21, 2024. Meredith Ringel Morris and coauthors propose “Levels of AGI,” a framework for classifying the capabilities and behavior of AGI models and their precursors. They organize capability by performance and generality, and discuss how current systems fit the framework. The authors say the levels could provide a common language for comparing models, assessing risks, and measuring progress toward AGI. They also describe challenging requirements for future benchmarks to quantify systems’ behavior and capabilities against the levels. The framework connects capability to deployment considerations: the authors discuss how autonomy and risk interact with the levels and emphasize carefully selecting Human-AI Interaction paradigms for responsible and safe deployment. The abstract does not specify which benchmarks would meet those requirements or how the framework classifies particular current systems.","topicIds":["artificial-general-intelligence"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on October 28, 2026; originally published July 21, 2024. The available evidence is the publication page's abstract; it does not provide the framework's detailed levels or examples of how current systems fit them.","speaker":{"name":"Google DeepMind","roleAtSource":"Publisher"},"source":{"title":"Levels of AGI for Operationalizing Progress on the Path to AGI","publisher":"Google DeepMind","format":"Article","url":"https://deepmind.google/research/publications/66938/","publishedAt":"2024-07-21T00:00:00+00:00","dateBasis":"Date stated on the publisher’s page."},"evidence":[{"text":"Meredith Ringel Morris and coauthors propose a framework for classifying AGI models and their precursors by performance, generality, and autonomy.","excerpt":"We propose a framework for classifying the capabilities and behavior of Artificial General Intelligence (AGI) models and their precursors. This framework introduces levels of AGI performance, generality, and autonomy","url":"https://deepmind.google/research/publications/66938/","relationship":"Source-stated"},{"text":"The authors say future benchmarks face challenging requirements to quantify model behavior and capabilities against the levels.","excerpt":"We discuss the challenging requirements for future benchmarks that quantify the behavior and capabilities of AGI models against these levels.","url":"https://deepmind.google/research/publications/66938/","relationship":"Source-stated"},{"text":"The authors connect the levels to deployment considerations, including autonomy and risk, and emphasize careful selection of Human-AI Interaction paradigms.","excerpt":"Finally, we discuss how these levels of AGI interact with deployment considerations such as autonomy and risk, and emphasize the importance of carefully selecting Human-AI Interaction paradigms for responsible and safe deployment of highly capable AI systems.","url":"https://deepmind.google/research/publications/66938/","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://deepmind.google/research/publications/66938/","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What benchmark designs could quantify model behavior and capabilities against the proposed levels?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Meredith Ringel Morris and coauthors propose an AGI framework at Google DeepMind","summary":"The paper was published on July 21, 2024. Meredith Ringel Morris and coauthors propose “Levels of AGI,” a framework for classifying the capabilities and behavior of AGI models and their precursors. They organize capability by performance and generality, and discuss how current systems fit the framework. The authors say the levels could provide a common language for comparing models, assessing risks, and measuring progress toward AGI. They also describe challenging requirements for future benchmarks to quantify systems’ behavior and capabilities against the levels. The framework connects capability to deployment considerations: the authors discuss how autonomy and risk interact with the levels and emphasize carefully selecting Human-AI Interaction paradigms for responsible and safe deployment. The abstract does not specify which benchmarks would meet those requirements or how the framework classifies particular current systems.","question":"What benchmark designs could quantify model behavior and capabilities against the proposed levels?","claims":[{"text":"Meredith Ringel Morris and coauthors propose a framework for classifying AGI models and their precursors by performance, generality, and autonomy.","excerpt":"We propose a framework for classifying the capabilities and behavior of Artificial General Intelligence (AGI) models and their precursors. This framework introduces levels of AGI performance, generality, and autonomy","support":"source-stated"},{"text":"The authors say future benchmarks face challenging requirements to quantify model behavior and capabilities against the levels.","excerpt":"We discuss the challenging requirements for future benchmarks that quantify the behavior and capabilities of AGI models against these levels.","support":"source-stated"},{"text":"The authors connect the levels to deployment considerations, including autonomy and risk, and emphasize careful selection of Human-AI Interaction paradigms.","excerpt":"Finally, we discuss how these levels of AGI interact with deployment considerations such as autonomy and risk, and emphasize the importance of carefully selecting Human-AI Interaction paradigms for responsible and safe deployment of highly capable AI systems.","support":"source-stated"}],"limitations":["Text as retrieved on October 28, 2026; originally published July 21, 2024.","The available evidence is the publication page's abstract; it does not provide the framework's detailed levels or examples of how current systems fit them."],"mediaId":null,"correctionReason":null,"id":"89333371ed8a8bbc2619ab81611cd65c2701b02df31817bd0f51b8f4b41faeaf","publicationId":"874fdcb1-69ec-c10f-ea5c-3f9283271205","version":1,"sourceUrl":"https://deepmind.google/research/publications/66938/","sourceTitle":"Levels of AGI for Operationalizing Progress on the Path to AGI","publisher":"Google DeepMind","publishedAt":"2024-07-21T00:00:00+00:00","originalPublished":{"value":"2024-07-21T00:00:00+00:00","precision":"instant","basis":"backfill:article_published"},"firstObservedAt":"2026-10-28T07:39:56.365Z","createdAt":"2026-11-17T17:24:00.554Z","historical":true,"entityIds":["google-deepmind"],"topicIds":["artificial-general-intelligence"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://deepmind.google/research/publications/66938/","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"07846da93df64867a88446e29290c0a55141c275e72730e9ab09fb26a7ad9327","rubricVersion":"unrecorded","provider":"typesafe","model":"jev-1.13.0","reasonCodes":["bridge_admitted_by_model"],"confidence":0.82,"decidedAt":"2026-10-28T07:39:55.136Z"},"provenance":{"textHash":"ba2312eb7d8e6bb2886439f945659c048a2443281e7d3b129335e8c5a4c254df","observationId":"7422ad8eae23d56068d13586376db943f797ba131e993c6feea8113474a47a45","retrievedAt":"2026-10-28T07:41:48.620Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"8a2268f7-ff56-25c1-3e41-724d8e4fd566"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"39d870e4-c862-c9f6-5d3c-cd027c11be14"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-17T17:19:26.459Z","model":"anthropic/claude-opus-5.5"}}}],"audits":[{"publicationId":"874fdcb1-69ec-c10f-ea5c-3f9283271205","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading attributes the framework to Meredith Ringel Morris and coauthors, who are the listed authors, on a Google DeepMind publication page. It presents the proposals as the authors' own and does not adopt them as established fact.","excerpt":"Meredith Ringel Morris, Jascha Sohl-dickstein, Noah Fiedel, Tris Warkentin, Allan Dafoe, Aleksandra Faust, Clement Farbaret, Shane Legg"},{"dimension":"summary","status":"pass","explanation":"The summary faithfully tracks the abstract: classification by performance, generality and autonomy; a common language for comparing models; benchmark requirements; and links to deployment, autonomy and risk. It states plainly what the abstract omits. AGI and benchmark are defined in the term library. The question follows from the abstract's stated benchmark challenge and is not leading.","excerpt":"providing a common language to compare models, assess risks, and measure progress along the path to AGI"},{"dimension":"date","status":"uncertain","explanation":"The reading states the publication date as July 21, 2024, matching the page dateline. However, that date may mark the ICML 2024 venue version rather than the work's first public release. The page does not show an earlier submission date, so this cannot be confirmed. The page was also modified after publication (December 2025). The reading presents no hindsight framing or later events.","excerpt":"July 21, 2024"},{"dimension":"support","status":"pass","explanation":"All three claims are source-stated and backed by verbatim abstract text. The page was edited after publication, but the cited abstract is plainly part of the original paper. The summary adds no unsupported benefits, costs or forecasts.","excerpt":"We discuss the challenging requirements for future benchmarks that quantify the behavior and capabilities of AGI models against these levels."},{"dimension":"media","status":"pass","explanation":"No media is selected and none was supplied or verified, so there is nothing to misidentify.","excerpt":null}],"createdAt":"2026-11-17T17:19:26.459Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}