{"id":"7ceaf834a930b0a33d62925c8bcc2f12d91a59cb17271d313bea6e7f57c57fa0","title":"arXiv q-bio.NC: CTM-AI, a Multi-Agent Intelligence Blueprint Inspired by Consciousness","argument":"Authors Haofei Yu, Yining Zhao, Lenore Blum, Manuel Blum and Paul Pu Liang propose CTM-AI, a multi-agent AI system inspired by the Conscious Turing Machine (CTM), a formal machine model of consciousness. Rather than treating consciousness as a demonstrated property, they present the model as a blueprint for coordinating many kinds of processors, including specialized experts and general-purpose learners. Their proposed mechanism competitively selects and integrates information from processors, while feedback lets the system change its internal connections and communication. The authors report strong results on MUStARD and UR-FUNNY, and improvements on tool-using and agentic tasks. These results describe performance on named benchmarks; the abstract does not establish how well CTM-AI works in other settings or whether it has human-like generality. The paper’s central idea is to make a model inspired by consciousness into a testable design for coordinating AI capabilities. How does CTM-AI perform beyond the benchmarks reported in the abstract?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the arXiv page and abstract, not the full paper; it does not provide evaluation methods, comparison details, or results beyond those named in the abstract. The reported benchmark results do not establish performance in other settings, human-like generality, or consciousness.","speaker":{"name":"Yu, Haofei, Zhao, Yining, Blum, Lenore, Blum, Manuel, Liang, Paul Pu","roleAtSource":"Article author"},"source":{"title":"CTM-AI: Evolving Multi-Agent Intelligence Inspired by a Model of Consciousness","publisher":"arXiv q-bio.NC","format":"Article","url":"https://arxiv.org/abs/2605.04097","publishedAt":"2026-04-30","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors present CTM-AI as a multi-agent AI system inspired by the Conscious Turing Machine, a formal machine model of consciousness.","excerpt":"we present a concrete blueprint for a multi-agent AI system inspired by a formal machine model of consciousness called the Conscious Turing Machine (CTM).","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"},{"text":"The system combines specialized processors with general-purpose learners that may develop expertise.","excerpt":"a large collection of powerful processors ranging from specialized experts (e.g., language models, multimodal models, tools, and APIs) to unspecialized general-purpose learners poised to develop their own expertise.","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"},{"text":"The authors describe selecting and integrating processor information, with feedback enabling changes to internal connections and communication.","excerpt":"information from many processors is competitively selected, integrated, and exchanged appropriately to solve the task. Feedback mechanisms in CTM-AI enable the system to iteratively evolve its internal connections and communication.","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"},{"text":"The authors report state-of-the-art accuracy on MUStARD and UR-FUNNY, with scores of 72.28 and 72.13.","excerpt":"CTM-AI achieves state-of-the-art accuracy on MUStARD (72.28) and UR-FUNNY (72.13)","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"},{"text":"The authors report improvements of more than 10 points on StableToolBench and WebArena-Lite for tool-using and agentic tasks.","excerpt":"On tool-using and agentic tasks, CTM-AI achieves 10+ points of improvement on StableToolBench and WebArena-Lite.","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2605.04097","checkedAt":"2026-11-03T20:24:59.074Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How does CTM-AI perform beyond the benchmarks reported in the abstract?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv q-bio.NC: CTM-AI, a Multi-Agent Intelligence Blueprint Inspired by Consciousness","summary":"Authors Haofei Yu, Yining Zhao, Lenore Blum, Manuel Blum and Paul Pu Liang propose CTM-AI, a multi-agent AI system inspired by the Conscious Turing Machine (CTM), a formal machine model of consciousness. Rather than treating consciousness as a demonstrated property, they present the model as a blueprint for coordinating many kinds of processors, including specialized experts and general-purpose learners. Their proposed mechanism competitively selects and integrates information from processors, while feedback lets the system change its internal connections and communication. The authors report strong results on MUStARD and UR-FUNNY, and improvements on tool-using and agentic tasks. These results describe performance on named benchmarks; the abstract does not establish how well CTM-AI works in other settings or whether it has human-like generality. The paper’s central idea is to make a model inspired by consciousness into a testable design for coordinating AI capabilities. How does CTM-AI perform beyond the benchmarks reported in the abstract?","question":"How does CTM-AI perform beyond the benchmarks reported in the abstract?","claims":[{"text":"The authors present CTM-AI as a multi-agent AI system inspired by the Conscious Turing Machine, a formal machine model of consciousness.","excerpt":"we present a concrete blueprint for a multi-agent AI system inspired by a formal machine model of consciousness called the Conscious Turing Machine (CTM).","support":"source-stated"},{"text":"The system combines specialized processors with general-purpose learners that may develop expertise.","excerpt":"a large collection of powerful processors ranging from specialized experts (e.g., language models, multimodal models, tools, and APIs) to unspecialized general-purpose learners poised to develop their own expertise.","support":"source-stated"},{"text":"The authors describe selecting and integrating processor information, with feedback enabling changes to internal connections and communication.","excerpt":"information from many processors is competitively selected, integrated, and exchanged appropriately to solve the task. Feedback mechanisms in CTM-AI enable the system to iteratively evolve its internal connections and communication.","support":"source-stated"},{"text":"The authors report state-of-the-art accuracy on MUStARD and UR-FUNNY, with scores of 72.28 and 72.13.","excerpt":"CTM-AI achieves state-of-the-art accuracy on MUStARD (72.28) and UR-FUNNY (72.13)","support":"source-stated"},{"text":"The authors report improvements of more than 10 points on StableToolBench and WebArena-Lite for tool-using and agentic tasks.","excerpt":"On tool-using and agentic tasks, CTM-AI achieves 10+ points of improvement on StableToolBench and WebArena-Lite.","support":"source-stated"}],"limitations":["The supplied evidence is the arXiv page and abstract, not the full paper; it does not provide evaluation methods, comparison details, or results beyond those named in the abstract.","The reported benchmark results do not establish performance in other settings, human-like generality, or consciousness."],"mediaId":"9b9bebfd703d0e801375f88b0defd942d2d36e20f3dd972dadc15a2a34201144","correctionReason":null,"id":"7ceaf834a930b0a33d62925c8bcc2f12d91a59cb17271d313bea6e7f57c57fa0","publicationId":"701ad398-234b-4689-8d7f-9931d3810621","version":1,"sourceUrl":"https://arxiv.org/abs/2605.04097","sourceTitle":"CTM-AI: Evolving Multi-Agent Intelligence Inspired by a Model of Consciousness","publisher":"arXiv q-bio.NC","publishedAt":null,"originalPublished":{"value":"2026-04-30","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-03T08:51:18.670Z","createdAt":"2026-11-03T18:17:28.990Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"9b9bebfd703d0e801375f88b0defd942d2d36e20f3dd972dadc15a2a34201144","kind":"article","url":"https://arxiv.org/abs/2605.04097","sourceUrl":"https://arxiv.org/abs/2605.04097","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-03T18:16:34.976Z","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/2605.04097","checkedAt":"2026-11-03T20:24:59.074Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"9b9bebfd703d0e801375f88b0defd942d2d36e20f3dd972dadc15a2a34201144","confidence":0.93,"provider":"typesafe","model":"jev-1.13.0","inputHash":"e3a9161551951426477bb930faa9d192b0e0241981edb2fd88dd2632eebd4fb3","decidedAt":"2026-11-03T18:16:36.238Z"},"provenance":{"textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observationId":"736b3a158a67aec006516fcc451b0329742a56aec9526d05422f169c3cc1f78f","retrievedAt":"2026-11-03T18:16:34.789Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"26effd00-b80e-411f-829d-8bb724bec951"},"reviewer":{"model":"gpt-6-luna","runId":"62bae3ad-33b3-437c-8ee4-57eb57922bb2"}},"audit":{"status":"passed","reviewedAt":"2026-11-03T18:17:28.990Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Yu, Haofei","url":null,"kind":"unknown"},{"name":"Zhao, Yining","url":null,"kind":"unknown"},{"name":"Blum, Lenore","url":null,"kind":"unknown"},{"name":"Blum, Manuel","url":null,"kind":"unknown"},{"name":"Liang, Paul Pu","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Yu, Haofei","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yu, Haofei","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}},{"actor":{"name":"Zhao, Yining","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhao, Yining","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}},{"actor":{"name":"Blum, Lenore","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Blum, Lenore","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}},{"actor":{"name":"Blum, Manuel","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Blum, Manuel","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}},{"actor":{"name":"Liang, Paul Pu","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liang, Paul Pu","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2605.04097","snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z","basis":"publisher-page-metadata","sourceLinks":47,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/7ceaf834a930b0a33d62925c8bcc2f12d91a59cb17271d313bea6e7f57c57fa0/source"}},"canonicalUrl":"https://escapemeridian.com/readings/arxiv-q-bio-nc-ctm-ai-a-multi-agent-intelligence-blueprint-7ceaf834","formats":{"html":"/readings/arxiv-q-bio-nc-ctm-ai-a-multi-agent-intelligence-blueprint-7ceaf834","markdown":"/readings/arxiv-q-bio-nc-ctm-ai-a-multi-agent-intelligence-blueprint-7ceaf834.md","json":"/readings/arxiv-q-bio-nc-ctm-ai-a-multi-agent-intelligence-blueprint-7ceaf834.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-agentic","name":"Agentic AI","explanation":"AI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.","referenceUrl":"https://developers.google.com/machine-learning/glossary#agentic"},{"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":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"7ceaf834a930b0a33d62925c8bcc2f12d91a59cb17271d313bea6e7f57c57fa0","revision":"701ad398-234b-4689-8d7f-9931d3810621","revisionKind":"publication-id","title":"arXiv q-bio.NC: CTM-AI, a Multi-Agent Intelligence Blueprint Inspired by Consciousness","url":"https://escapemeridian.com/readings/arxiv-q-bio-nc-ctm-ai-a-multi-agent-intelligence-blueprint-7ceaf834"},"source":{"type":"source","id":"https://arxiv.org/abs/2605.04097","revision":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","revisionKind":"sha256-source-text","title":"CTM-AI: Evolving Multi-Agent Intelligence Inspired by a Model of Consciousness","url":"https://arxiv.org/abs/2605.04097"},"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":"7ceaf834a930b0a33d62925c8bcc2f12d91a59cb17271d313bea6e7f57c57fa0","title":"arXiv q-bio.NC: CTM-AI, a Multi-Agent Intelligence Blueprint Inspired by Consciousness","argument":"Authors Haofei Yu, Yining Zhao, Lenore Blum, Manuel Blum and Paul Pu Liang propose CTM-AI, a multi-agent AI system inspired by the Conscious Turing Machine (CTM), a formal machine model of consciousness. Rather than treating consciousness as a demonstrated property, they present the model as a blueprint for coordinating many kinds of processors, including specialized experts and general-purpose learners. Their proposed mechanism competitively selects and integrates information from processors, while feedback lets the system change its internal connections and communication. The authors report strong results on MUStARD and UR-FUNNY, and improvements on tool-using and agentic tasks. These results describe performance on named benchmarks; the abstract does not establish how well CTM-AI works in other settings or whether it has human-like generality. The paper’s central idea is to make a model inspired by consciousness into a testable design for coordinating AI capabilities. How does CTM-AI perform beyond the benchmarks reported in the abstract?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the arXiv page and abstract, not the full paper; it does not provide evaluation methods, comparison details, or results beyond those named in the abstract. The reported benchmark results do not establish performance in other settings, human-like generality, or consciousness.","speaker":{"name":"Yu, Haofei, Zhao, Yining, Blum, Lenore, Blum, Manuel, Liang, Paul Pu","roleAtSource":"Article author"},"source":{"title":"CTM-AI: Evolving Multi-Agent Intelligence Inspired by a Model of Consciousness","publisher":"arXiv q-bio.NC","format":"Article","url":"https://arxiv.org/abs/2605.04097","publishedAt":"2026-04-30","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors present CTM-AI as a multi-agent AI system inspired by the Conscious Turing Machine, a formal machine model of consciousness.","excerpt":"we present a concrete blueprint for a multi-agent AI system inspired by a formal machine model of consciousness called the Conscious Turing Machine (CTM).","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"},{"text":"The system combines specialized processors with general-purpose learners that may develop expertise.","excerpt":"a large collection of powerful processors ranging from specialized experts (e.g., language models, multimodal models, tools, and APIs) to unspecialized general-purpose learners poised to develop their own expertise.","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"},{"text":"The authors describe selecting and integrating processor information, with feedback enabling changes to internal connections and communication.","excerpt":"information from many processors is competitively selected, integrated, and exchanged appropriately to solve the task. Feedback mechanisms in CTM-AI enable the system to iteratively evolve its internal connections and communication.","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"},{"text":"The authors report state-of-the-art accuracy on MUStARD and UR-FUNNY, with scores of 72.28 and 72.13.","excerpt":"CTM-AI achieves state-of-the-art accuracy on MUStARD (72.28) and UR-FUNNY (72.13)","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"},{"text":"The authors report improvements of more than 10 points on StableToolBench and WebArena-Lite for tool-using and agentic tasks.","excerpt":"On tool-using and agentic tasks, CTM-AI achieves 10+ points of improvement on StableToolBench and WebArena-Lite.","url":"https://arxiv.org/abs/2605.04097","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2605.04097","checkedAt":"2026-11-03T20:24:59.074Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How does CTM-AI perform beyond the benchmarks reported in the abstract?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv q-bio.NC: CTM-AI, a Multi-Agent Intelligence Blueprint Inspired by Consciousness","summary":"Authors Haofei Yu, Yining Zhao, Lenore Blum, Manuel Blum and Paul Pu Liang propose CTM-AI, a multi-agent AI system inspired by the Conscious Turing Machine (CTM), a formal machine model of consciousness. Rather than treating consciousness as a demonstrated property, they present the model as a blueprint for coordinating many kinds of processors, including specialized experts and general-purpose learners. Their proposed mechanism competitively selects and integrates information from processors, while feedback lets the system change its internal connections and communication. The authors report strong results on MUStARD and UR-FUNNY, and improvements on tool-using and agentic tasks. These results describe performance on named benchmarks; the abstract does not establish how well CTM-AI works in other settings or whether it has human-like generality. The paper’s central idea is to make a model inspired by consciousness into a testable design for coordinating AI capabilities. How does CTM-AI perform beyond the benchmarks reported in the abstract?","question":"How does CTM-AI perform beyond the benchmarks reported in the abstract?","claims":[{"text":"The authors present CTM-AI as a multi-agent AI system inspired by the Conscious Turing Machine, a formal machine model of consciousness.","excerpt":"we present a concrete blueprint for a multi-agent AI system inspired by a formal machine model of consciousness called the Conscious Turing Machine (CTM).","support":"source-stated"},{"text":"The system combines specialized processors with general-purpose learners that may develop expertise.","excerpt":"a large collection of powerful processors ranging from specialized experts (e.g., language models, multimodal models, tools, and APIs) to unspecialized general-purpose learners poised to develop their own expertise.","support":"source-stated"},{"text":"The authors describe selecting and integrating processor information, with feedback enabling changes to internal connections and communication.","excerpt":"information from many processors is competitively selected, integrated, and exchanged appropriately to solve the task. Feedback mechanisms in CTM-AI enable the system to iteratively evolve its internal connections and communication.","support":"source-stated"},{"text":"The authors report state-of-the-art accuracy on MUStARD and UR-FUNNY, with scores of 72.28 and 72.13.","excerpt":"CTM-AI achieves state-of-the-art accuracy on MUStARD (72.28) and UR-FUNNY (72.13)","support":"source-stated"},{"text":"The authors report improvements of more than 10 points on StableToolBench and WebArena-Lite for tool-using and agentic tasks.","excerpt":"On tool-using and agentic tasks, CTM-AI achieves 10+ points of improvement on StableToolBench and WebArena-Lite.","support":"source-stated"}],"limitations":["The supplied evidence is the arXiv page and abstract, not the full paper; it does not provide evaluation methods, comparison details, or results beyond those named in the abstract.","The reported benchmark results do not establish performance in other settings, human-like generality, or consciousness."],"mediaId":"9b9bebfd703d0e801375f88b0defd942d2d36e20f3dd972dadc15a2a34201144","correctionReason":null,"id":"7ceaf834a930b0a33d62925c8bcc2f12d91a59cb17271d313bea6e7f57c57fa0","publicationId":"701ad398-234b-4689-8d7f-9931d3810621","version":1,"sourceUrl":"https://arxiv.org/abs/2605.04097","sourceTitle":"CTM-AI: Evolving Multi-Agent Intelligence Inspired by a Model of Consciousness","publisher":"arXiv q-bio.NC","publishedAt":null,"originalPublished":{"value":"2026-04-30","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-03T08:51:18.670Z","createdAt":"2026-11-03T18:17:28.990Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"9b9bebfd703d0e801375f88b0defd942d2d36e20f3dd972dadc15a2a34201144","kind":"article","url":"https://arxiv.org/abs/2605.04097","sourceUrl":"https://arxiv.org/abs/2605.04097","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-03T18:16:34.976Z","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/2605.04097","checkedAt":"2026-11-03T20:24:59.074Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"9b9bebfd703d0e801375f88b0defd942d2d36e20f3dd972dadc15a2a34201144","confidence":0.93,"provider":"typesafe","model":"jev-1.13.0","inputHash":"e3a9161551951426477bb930faa9d192b0e0241981edb2fd88dd2632eebd4fb3","decidedAt":"2026-11-03T18:16:36.238Z"},"provenance":{"textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observationId":"736b3a158a67aec006516fcc451b0329742a56aec9526d05422f169c3cc1f78f","retrievedAt":"2026-11-03T18:16:34.789Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"26effd00-b80e-411f-829d-8bb724bec951"},"reviewer":{"model":"gpt-6-luna","runId":"62bae3ad-33b3-437c-8ee4-57eb57922bb2"}},"audit":{"status":"passed","reviewedAt":"2026-11-03T18:17:28.990Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Yu, Haofei","url":null,"kind":"unknown"},{"name":"Zhao, Yining","url":null,"kind":"unknown"},{"name":"Blum, Lenore","url":null,"kind":"unknown"},{"name":"Blum, Manuel","url":null,"kind":"unknown"},{"name":"Liang, Paul Pu","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Yu, Haofei","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yu, Haofei","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}},{"actor":{"name":"Zhao, Yining","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhao, Yining","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}},{"actor":{"name":"Blum, Lenore","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Blum, Lenore","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}},{"actor":{"name":"Blum, Manuel","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Blum, Manuel","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}},{"actor":{"name":"Liang, Paul Pu","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liang, Paul Pu","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.04097","locator":"head/meta[name=author|citation_author]"},"assertionId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-03T20:24:59.074Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","sourceUrl":"https://arxiv.org/abs/2605.04097","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2605.04097","snapshotId":"6f1f46092f3e8a002d02b4f0a0ea7d8555488d388b4ae4ad4d5851679a25aa5d","contentSha256":"890f471dde846d7a190f1d3bb60ed72849fa2ab5f88c83dfa3bc7a7f2554d4af","textHash":"60f8b487f9f8dedbf7108c1af4a80c606c16936f1a2cb9ffbc246a318676bab6","observedAt":"2026-11-03T20:24:59.074Z","basis":"publisher-page-metadata","sourceLinks":47,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/7ceaf834a930b0a33d62925c8bcc2f12d91a59cb17271d313bea6e7f57c57fa0/source"}}}],"audits":[{"publicationId":"701ad398-234b-4689-8d7f-9931d3810621","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The draft names the five listed authors (Haofei Yu, Yining Zhao, Lenore Blum, Manuel Blum, Paul Pu Liang). It keeps the design, the benchmark scores and the improvement claims attributed to them as authors' reports, not as established fact.","excerpt":"Authors:Haofei Yu, Yining Zhao, Lenore Blum, Manuel Blum, Paul Pu Liang"},{"dimension":"summary","status":"pass","explanation":"The draft leads with the actor and the proposal, and explains CTM in plain terms. It describes the mechanism of competitive selection, integration and feedback, and keeps the authors' 'flipping the debate' framing without claiming the system is conscious. MUStARD, UR-FUNNY, StableToolBench and WebArena-Lite are not explained, but the abstract does not define them. Calling them named benchmarks is enough for a general reader. Optional repair: add term-library tooltips for these benchmarks and for 'agentic'. The question follows from the evidence and is not leading.","excerpt":"Flipping the debate, we present a concrete blueprint for a multi-agent AI system inspired by a formal machine model of consciousness called the Conscious Turing Machine (CTM)."},{"dimension":"date","status":"pass","explanation":"The displayed original publication date of 2026-04-30 matches the v1 submission date. It does not use the v2 revision of 1 Nov 2026 or the November feed date.","excerpt":"[Submitted on 30 Apr 2026 (v1), last revised 1 Nov 2026 (this version, v2)]"},{"dimension":"support","status":"pass","explanation":"Each claim matches its quoted abstract text, including the scores 72.28 and 72.13 and the 10+ point gains. The caveats do not overreach: benchmark results are not real-world reliability, and the abstract alone does not show generality or consciousness. 'Testable design' reflects the authors' own wording. No benefits, costs or stakeholders are invented.","excerpt":"Overall, CTM-AI offers a principled and testable multi-agent AI system inspired by a model of consciousness."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct link to the original arXiv abstract page. It matches the source URL and is the most direct way to reach the source. No better source media was found.","excerpt":null}],"createdAt":"2026-11-03T18:17:28.990Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}