{"resourceId":"pnnl-emotr-ai-em-summary-202512","versions":[{"version":"external-42a6691ed728cb8fb64026df7b292dff5dabcafff32b3fcf0f348c9ac6e8bbdc","resource":{"id":"pnnl-emotr-ai-em-summary-202512","title":"PNNL EMOTR summary: EOC AI hinges on data interoperability and human readiness, not models alone","organization":"Pacific Northwest National Laboratory (for DHS Science and Technology Directorate)","sector":"Federal laboratory research / emergency management AI","geography":"United States (DHS S&T EM stakeholder engagement)","publishedAt":"2025-12","publicationDate":"2025-12-01","eventDate":null,"sourceName":"Pacific Northwest National Laboratory","sourceLabel":"PNNL-38842 EMOTR Summary Report: Leveraging AI for Emergency Management","sourceUrl":"https://www.pnnl.gov/sites/default/files/media/file/PNNL-38842_EMOTR_Summary_Report.pdf","evidenceClass":"government-evaluation","outcomeClass":"mixed","topics":["operating-model","infrastructure","governance-procurement","accessibility-workforce","data-security"],"finding":"PNNL-38842 (Emergency Management of Tomorrow Research Summary Report, December 2025; PDF metadata CreationDate 2026-09-22) synthesizes DHS S&T–PNNL work on AI-enabled EM and EOCs. The executive summary states that lacking effective data management for situational awareness, interoperability, and common operating pictures hampers EM communication, sharing, allocation, and real-time decisions. The report profiles infrastructure, agent protocols, tooling/enrichment, memory/personalization, and governance/operations approaches, and catalogs adoption barriers including culture/training, trust/explainability, and the need for high human readiness alongside technology readiness—noting infrequent events limit familiarity with AI tools used during incidents.","sledRelevance":"Primary federal lab synthesis for SLTT EM offices designing AI roadmaps beside tools like LINK. Why now: PDF CreationDate 2026-09-22 makes this previously unarchived EMOTR summary newly available even though the report month is December 2025, and it contextualizes the Sept 23 LINK demo’s long transition path.","evidence":"Opened PNNL PDF via curl/pdftotext. Cover identity PNNL-38842, December 2025 dating, executive-summary data-management statement, technology-profile list, and human-readiness/infrequent-event barrier language taken from the report; pdfinfo CreationDate 2026-09-22 recorded for freshness.","architectureImplications":"Prioritize data access architectures, provenance/trust frameworks, and agent/tool protocols before stacking generative assistants in the EOC.","governanceImplications":"Define AI/human roles and usage conditions (HRL) before deployment; treat vendor demos as incomplete without workflow analysis.","securityPrivacyImplications":"Multi-jurisdictional knowledge access, data sovereignty, and provenance controls are first-class requirements for any cross-EOC AI sharing.","caveats":"Framework synthesis, not a randomized outcome study; some barrier citations are secondary surveys; report month Dec 2025 with Sept 2026 PDF packaging.","streamIds":["emergency-services"],"roles":{"sales":"Position EMOTR as cautionary federal guidance: EOC AI value is gated by data interoperability, provenance, and human readiness—not model novelty. Discovery should inventory COP data feeds, cross-jurisdiction sharing limits, and whether staff rehearse AI tools between incidents. Bounded engagement: EM AI readiness assessment covering data architecture and HRL/TRL gaps. Unsupported claims include guaranteed real-time COP AI, autonomous agent operations in disasters, or that buying an LLM fixes interoperability.","engineering":"Fit is an architecture and governance blueprint for AI-enabled EM data/EOC systems—profiles spanning storage/access, agent protocols, tooling, memory under missing data, and governance—not a single product. Integrate provenance, multi-jurisdictional access controls, and human-on-the-loop interfaces before agentic automation. Prerequisites: characterized data processes, offline/degraded-mode plans, and role definitions. Constraints: generative AI compute/energy demands; infrequent-event unfamiliarity. Proof of value is improved shared COP completeness and decision latency with humans still accountable.","delivery":"Delivery should sequence data-interop and training work before pilot agents, with tabletop exercises that force AI use under stress. Dependencies include DHS/S&T guidance, EOC IT, and partner agencies’ data-sharing agreements. Ownership: EM director for operational policy; CIO/GIS for platforms. Acceptance criteria: documented AI/human role matrix, provenance logging, and rehearsal evidence. Risks include shelfware after rare events and cross-border data-sovereignty violations."},"roleDetails":{"sales":{"customerProblem":"EOCs want AI for situational awareness but lack interoperable data and practiced human-AI workflows.","stakeholders":["EM directors","EOC managers","State fusion/GIS leads","DHS/FEMA partners"],"discoveryQuestions":["Which COP data sources still require manual swivel-chair?","Do staff rehearse AI tools between incidents?","What cross-jurisdiction data-sovereignty rules apply?","Are AI vs human roles written into SOPs?"],"valueHypothesis":"AI helps EM when data interoperability and human readiness are funded alongside models—not when models arrive alone.","potentialEngagement":"EM AI readiness and data-architecture assessment mapped to EMOTR profiles.","unsupportedClaims":["LLM purchase alone fixes COP gaps","Autonomous disaster agents ready for EOC command","Guaranteed real-time multi-jurisdiction AI sharing"]},"engineering":{"fit":"Reference architecture for AI-enabled EM data/EOC systems emphasizing interoperability and human-on-the-loop control.","architectureIntegration":"Data access/provenance layers → optional agent/tool protocols → human-validated COP enrichment with degraded-mode paths.","prerequisites":["Data-process characterization","Provenance/trust framework","HRL/TRL role definitions","Offline/degraded plans"],"constraints":["Infrequent-event unfamiliarity","Compute/energy for generative AI","Multi-jurisdiction compliance"],"security":"Enforce data sovereignty, least privilege on vulnerability-rich infrastructure data, and audit of AI-enriched COP products.","proofOfValue":["COP completeness deltas","Decision-latency with HITL","Rehearsal compliance rates"]},"delivery":{"work":"Stand up interop/provenance baselines, write AI/human SOPs, and run stress tabletops before expanding agents.","dependencies":["EOC IT/GIS","Partner data-sharing MOUs","Training cadre","S&T/PNNL guidance updates"],"ownership":"EM director owns policy; platform owner owns technical controls.","skillsAdoption":"Train all EOC roles—not only specialists—on when to use or ignore AI under stress.","governanceCheckpoints":["HRL/TRL gate","Data-sharing legal review","Post-exercise after-action"],"acceptanceCriteria":["Role matrix signed","Provenance logs live","Documented rehearsal of AI workflows"],"risks":["Shelfware after rare events","Sovereignty breaches","Automation complacency"]}},"retrievedAt":"2026-09-26T22:03:08-05:00","enrichedAt":"2026-09-26T22:03:08-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Training and human-readiness gaps—not just model choice—determine whether AI is used or abandoned under stress.","procurementImplications":"Procure data interoperability, provenance, and training packages with any EM AI tool; avoid model-only buys.","operatingModelImplications":"Practice AI-enabled workflows between incidents so infrequent disasters do not debut unfamiliar tools.","sourceVerification":{"openedUrl":"https://www.pnnl.gov/sites/default/files/media/file/PNNL-38842_EMOTR_Summary_Report.pdf","referenceExcerpt":"A lack of effective data management approaches for situational awareness, interoperability, and integration","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}