A public-health agency plans to pilot a GenAI triage-support tool across several divisions, but the readiness review finds that each division follows different, inconsistent record-keeping standards for patient intake data. What should the agency do before scaling the pilot?
Select an answer to reveal the explanation.
Short Explanation
Feeding a triage tool inconsistent intake records is like asking someone to read five patient charts, each written in a different shorthand; the output will be as reliable as the messiest input. Before this pilot scales, the agency needs one consistent record-keeping standard across divisions, not a tool clever enough to guess around the gaps.
Full Explanation
Data quality is a foundational readiness dimension because a GenAI tool prompted on inconsistent inputs produces inconsistent, and in a triage context potentially unsafe, outputs. When divisions follow different record-keeping standards, the same clinical situation can be documented in incompatible ways, and a tool asked to reason across that data cannot reliably tell a true difference in patient condition from a mere difference in formatting. Standardizing record-keeping before scaling closes that gap at the source, giving the tool a consistent foundation to reason from across every division it touches. Scaling to all divisions simultaneously multiplies the inconsistency rather than resolving it, and surfacing more errors faster is not the same as preventing them in a triage-support context where mistakes carry real clinical stakes. Limiting rollout to the most technically advanced division sidesteps the data problem instead of solving it, and the other divisions still need a fix before they can safely join. Trusting the GenAI tool's summarization ability to reconcile format differences overestimates what summarization does; it does not verify clinical accuracy across incompatible source data. Scope note: standardization does not require identical systems, only consistent fields and shared definitions. Operational check: audit a sample of intake records from each division against a shared minimum data standard before expanding the pilot.