{"resourceId":"ncdot-copilot-expectation-recalibration-2026","versions":[{"version":"external-aab1a3767aa79d68a882b3725be106eb7e199fea1952417b47db77ca4fad9dcb","resource":{"id":"ncdot-copilot-expectation-recalibration-2026","title":"NCDOT study finds staff expectations change after hands-on Copilot use","organization":"University of North Carolina at Charlotte and North Carolina Department of Transportation","sector":"State transportation workforce","geography":"North Carolina, United States","publishedAt":"July 15, 2026, arXiv v1; pilot conducted in 2025","publicationDate":"2026-07-15","eventDate":null,"sourceName":"Persona Migration and Expectation Recalibration in Generative AI Adoption: A Longitudinal Study at a State Department of Transportation","sourceLabel":"Academic preprint; university and agency coauthors","sourceUrl":"https://arxiv.org/html/2607.13798v1","evidenceClass":"academic-research","outcomeClass":"mixed","topics":["knowledge-work","infrastructure","data-security","governance-procurement","accessibility-workforce","operating-model"],"finding":"Perceived usefulness fell after the pilot; other aggregate acceptance constructs did not change significantly.","sledRelevance":"Direct state workforce evidence, relevant to shared productivity-suite rollout rather than traffic-system performance.","evidence":"Eight-week 2025 NCDOT pilot; 175 baseline respondents, 133 matched responses and 124 after screening. Table 3 reports usefulness 3.85 before versus 3.62 after, adjusted p<0.001. Paired survey analysis and exploratory clustering measure perceptions, not causal productivity effects.","architectureImplications":"Copilot operated in approved NCDOT Microsoft 365 applications. Interpretation: validate content permissions and task-specific integrations before license expansion.","governanceImplications":"Interpretation: collect recurring feedback while evaluating quality separately; enthusiasm is not an approval control.","securityPrivacyImplications":"Interpretation: retain access restrictions even when staff report greater confidence; verify permission and disclosure behavior directly.","caveats":"Single agency, short horizon, self-reports and no untreated comparison. Small persona transitions and keyword-based qualitative coding limit inference. Attrition checks cover measured baseline attitudes only. Preprint status; no verified net savings or accessibility effect.","streamIds":["state-government"],"roles":{"sales":"Interpretation: Ask the CIO, transportation business-unit leaders, learning team and budget owner which office tasks justify the license and what would count as sustained value. Use the study's changing usefulness ratings to discuss expectation setting, not to label staff resistant. A bounded engagement could assess a small set of communication, analysis and document workflows and recommend targeted enablement. The value hypothesis is more informed allocation of training and licenses. Ask how managers will measure quality and correction effort alongside perceived usefulness. Do not claim this survey proves productivity loss, financial return or improved transportation outcomes, and do not generalize its user clusters to another workforce.","engineering":"Interpretation: Begin with the existing collaboration tenant and identify precisely which documents and applications each test role may access. Prerequisites include approved test material, clear information ownership and accessible interaction paths. Validate representative drafting, summarization and spreadsheet tasks separately; an acceptable summary does not establish reliable numerical analysis. Instrument errors, source checking and correction time under a documented configuration. Include adversarial content in retrieved documents and tests for unauthorized information access. Compare observed task performance with the current process while separately collecting feedback. The research does not compare hosting architectures, coding assistants or autonomous agents, so those require their own technical assessment.","delivery":"Interpretation: The workforce enablement owner should coordinate business managers, platform IT, accessibility specialists and security. Deliver role-based examples, office hours and guidance on checking outputs, then repeat feedback collection after routine use. Dependencies include protected learning time, stable configuration and managers willing to hear disappointment. Proposed acceptance criteria: each selected workflow has an observed quality assessment and cost record, staff can demonstrate correction and escalation, and the license decision records unresolved fit gaps. Track adoption without treating high usage as success by itself. Risks include surveillance-like employee scoring, using preliminary personas as permanent labels and continuing subscriptions where local workflow evidence remains weak."},"retrievedAt":"2026-09-08T03:02:23Z","enrichedAt":"2026-09-08T03:06:00Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Interpretation: provide accessible, role-specific learning and confidential ways to report skill concerns without labeling individual employees.","procurementImplications":"Interpretation: plan license reassessment by demonstrated workflow fit and include vendor-change notice in commercial review.","operatingModelImplications":"Interpretation: workforce enablement and platform operations should jointly own recurring feedback and service improvement.","updateExplanation":"No matching source URL or arXiv identifier in the full archive or targeted search. Adds unarchived state-specific longitudinal evidence; July publication and 2025 pilot are explicitly historical.","sourceVerification":{"openedUrl":"https://arxiv.org/html/2607.13798v1","referenceExcerpt":"The study also reflects the context of a single state DOT.","promptVersion":"sled-research-v3.1","model":null,"basis":"agent-reported inspection"}}}]}