A city emergency AI incident bridge has network and facilities leads on call but no AI subject-matter expert. What should the auditor recommend?
Select an answer to reveal the explanation.
Short Explanation
When the model is the problem, you need someone who understands models on the bridge — not just a network lead authorizing containment. AI SME coverage belongs in the on-call roster. Infrastructure metrics alone won't debug the model, and deferring AI input to the next quarterly review is too late.
Full Explanation
Including AI expertise on municipal incident bridges for model-dependent systems works because containment choices—rollback versus retrain, threshold changes, prompt or tool-disable decisions, data quarantine—require people who understand model behavior, not only network paths and facilities status. Without AI SMEs, bridges may declare “systems green” while the model continues producing harmful civic outputs, or they may apply IT fixes that do not address decision failure modes. Auditors should expect named AI on-call coverage for services whose critical path includes models.
Assigning the on-call network lead to authorize model containment actions fails because network leads may lack the model behavior knowledge needed to assess drift, false alerts, or AI data integrity during emergencies. Routing municipal bridge escalation decisions through infrastructure health metrics fails conceptually: healthy network and facility metrics can mask ongoing model failures, so metrics alone cannot replace AI expertise for containment decisions. Scheduling AI subject-matter expert consultation for the next quarterly governance review meeting fails because role clarity and AI expertise are needed during the incident, not deferred to a future meeting when decisions have already been made without it.
Exam caveat: AI expertise can be vendor-shared or cross-department, but availability and authority must be pre-agreed; a name on a slide without paging rights is not bridge coverage. Operational check: update the city emergency AI bridge roster with primary/backup AI SMEs, define their decision rights, include them in IR paging trees, and run a tabletop where network and facilities leads must consult AI expertise before closing a model-impacting incident.