{"resourceId":"nc-treasurer-departmentwide-ai-rollout-20260413","versions":[{"version":"external-0e7afbba86479d267c38039c0ed2a938ae6d37defb5388de218152b424301ff1","resource":{"id":"nc-treasurer-departmentwide-ai-rollout-20260413","title":"Treasurer rollout moves evaluation responsibility to division-level operations","organization":"North Carolina Department of State Treasurer","sector":"State government","geography":"North Carolina, United States","publishedAt":"April 13, 2026","publicationDate":"2026-04-13","eventDate":null,"sourceName":"NC Treasurer","sourceLabel":"Official operator announcement; effectiveness claims are not independently evaluated","sourceUrl":"https://www.nctreasurer.gov/news/press-releases/2026/04/13/nc-department-state-treasurer-announces-extensive-implementation-artificial-intelligence","evidenceClass":"government-evaluation","outcomeClass":"emerging","topics":["knowledge-work","developers-agents","data-security","governance-procurement","accessibility-workforce","operating-model"],"finding":"The department reports broad AI implementation; replicated productivity gains remain an expectation.","sledRelevance":"Historical follow-through to the archived Treasurer pilot, relevant to state workforce scaling.","evidence":"The announcement describes multiple engines, staff training and IT tracking of use and effectiveness. It offers no department-wide sample, comparator or measured results.","architectureImplications":"Interpretation: map each approved engine to its workflow, identity boundary and data sources; evaluate code assistance separately from research.","governanceImplications":"Interpretation: require a division-specific continuation decision supported by quality and effort measures.","securityPrivacyImplications":"Interpretation: test data-loss controls and retention settings rather than treating policy statements as security evidence.","caveats":"Operator announcement, not an independent evaluation. Baseline methods, hosting, model versions and full operating costs are absent.","streamIds":["state-government"],"roles":{"sales":"Interpretation: For a finance agency expanding staff tools, the customer problem is determining where licenses and training produce useful work. Include division heads, IT, privacy, finance and workforce representatives. Ask which outputs require expert correction, whether staff have suitable tasks, and who pays for recurring support. A bounded engagement could assess one division through a baseline period and a limited assisted-work period. The value hypothesis is lower effort per accepted output at maintained quality. The rollout provides a reason to ask for continuing evidence; it cannot justify guaranteed savings, workforce reductions or transferring one team's claimed gains to every division.","engineering":"Interpretation: Fit is supervised assistance within approved tasks. Inventory engines, permissions, connectors, prompts and versions before integration. Prerequisites include representative tasks, reviewers and data classifications. Use synthetic sensitive records to test export restrictions and incorrect retrieval. For coding, require repository permissions, tests and review before merge; for research, require source verification. A proof of value should compare accepted outputs, correction time and total cost with the current workflow. Cloud, on-premises and hybrid deployment choices require local assessment because this announcement supplies no comparative evidence. Keep tools from autonomously changing financial or legal records during validation.","delivery":"Interpretation: Assign a division service owner to benefits measurement and IT to platform operation. Establish an approved task catalog, accessible training, support routes and a correction log. Dependencies include reviewer capacity, reliable access controls and records guidance. Train staff to decline unsuitable tasks and report errors without penalty. Proposed acceptance criteria: every sampled output has an accountable reviewer; all sensitive-data tests pass; time, quality and cost are reported against the agreed baseline before expansion. Review after material model changes. Risks include low adoption hidden by license counts and review work offsetting apparent drafting speed."},"retrievedAt":"2026-09-14T03:00:49Z","enrichedAt":"2026-09-14T03:02:46Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Interpretation: test assistive-technology compatibility and include paid learning and correction time in evaluation.","procurementImplications":"Interpretation: require task-specific evaluation access, change notices and export/exit terms for each engine.","operatingModelImplications":"Interpretation: division leaders own accepted outcomes; IT owns service reliability and access.","updateExplanation":"Absent from 274 full-archive records at offsets 0, 100 and 200; exact URL search returned zero. Adds the April department-wide transition to the separately archived pilot report; not a new September event.","sourceVerification":{"openedUrl":"https://www.nctreasurer.gov/news/press-releases/2026/04/13/nc-department-state-treasurer-announces-extensive-implementation-artificial-intelligence","referenceExcerpt":"Our information technology team will be tracking our AI use and effectiveness through this transition.","promptVersion":"sled-research-v3.2","model":null,"basis":"agent-reported inspection"}}}]}