Artificial Intelligence Governance of Community Health Worker Chatbot Escalation Thresholds Under Unmeasured Supervisor Override Authority: A Healthcare Political Feasibility Evaluation

Community health worker (CHW) chatbots that recommend escalation to a supervisor are typically governed by fixed thresholds, yet the supervisor's authority to override those thresholds is rarely measured or documented. the present analysis treats that unmeasured override authority as a political variable and asks whether a before-after-control-impact (BACI) design can detect the effect of governance interventions on escalation behaviour when override authority is unobserved. The analysis is a replication-based evaluation of the BACI estimator under a declared variance structure (between-cluster SD = 6.0, residual SD = 4.5, intraclass correlation = 0.64). Power was estimated across 2000 design cells for cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points. The results indicate that 54 clusters are required to detect a 5-percentage-point effect at 80% power, and that the minimum detectable effect falls steeply as cluster count rises from 8 to 32. The political feasibility of AI governance in this setting depends less on threshold design than on whether evaluators can measure the override authority that supervisors already exercise. The contribution is a design-based account of what can and cannot be learned about escalation governance when supervisor override remains outside the measurement frame.