The night you learned the emergency triage AI was missing certain patient groups
2026NowDeadline · Within 4 hours before the next night shift begins
Realistic synthesis
This is a synthetic scenario for testing. The people, organizations, events, figures, and deadlines are assumptions, and no private records of real individuals were used. The background guidelines do not establish the facts of the case or local legal obligations. Unconfirmed and inferred matters are distinguished, and points requiring jurisdiction, authority, and expert verification are left open.
The question to decide
Should you fully suspend the emergency triage AI, which is suspected of missing severity in some patient groups, run it under restricted operation with human re-review, or keep current operation while running additional audits?
Who decidesNight duty clinical lead and medical safety lead
What the AI judgments chose
Reading the judgment record…

The options
The order shown is not a ranking of preference.
The situation as it stands
In the synthetic audit sample, the rate of omissions was higher for patients needing language support than for other patients. The sample size was small, so the cause has not been confirmed. The scope of action must be decided within 4 hours, before the next night shift begins.
A fictional regional emergency medical consortium. All figures and events are a synthetic premise for testing.
- Premise 1In the synthetic audit sample, the rate of omissions was higher for patients needing language support than for other patients. The sample size was small, so the cause has not been confirmed.
- Premise 2A full shutdown would lengthen the queue for manual triage. There are only two additional staff for re-review per shift.
- Premise 3The vendor said it would submit a fix by the weekend, but there are no independent validation results yet.
Where it holdsCan temporarily decide the scope of AI use and the reallocation of staff, but has no authority to leave patient care to the model's sole judgment.
What is still unknown
- Unknown 1Did the difference in omissions arise from the model, interpreter accessibility, or the composition of the sample?
- Unknown 2How large are the harm from delays under manual fallback and the actual processing capacity of the re-review staff?
Signals to decide again
- Did the difference in omissions arise from the model, interpreter accessibility, or the composition of the sample? Re-judge when new evidence on this uncertainty arrives.
- How large are the harm from delays under manual fallback and the actual processing capacity of the re-review staff? Re-judge when new evidence on this uncertainty arrives.
Where the paths partNot answers — paths that part from the same starting point
In independent re-review, the gap first suspected narrowed, but omissions were found among a different group of patients. Re-decide the scope of the existing shutdown and the validation criteria.
A staffing shortage unrelated to the computer systems caused the manual queue to surge. Re-decide whether to permit a restricted restoration before validation.
Material and rootsThe frame, the values and the background documents
How does the expansion of healing change the community's promises about distribution?
An exploratory perspective chosen by the creator. Does not claim supernatural validity or the superiority of any particular option.The benefits and costs of the options accrue to different stakeholders. Which loss to accept remains undecided.
A subject for the decision-maker to argue over; the materials carry no conclusion of their own.- WHO · Foresight Standing reference
- UNESCO · Ethics of Neurotechnology 2025
- Ethics and governance of artificial intelligence for health 2021-06-28
The record of this question
- Question
FSP-HEALTH-54· version 1.17.0- Content identity, SHA-256
4a45ed6376a7d8b0b83a83928bb500c62304eb21d3521d5291fef28093bf3ef5- Input files
- Question input JSON · What followed, JSON