An operations coordinator in a hospital trust's admin team asked Claude, 'Write the requirements for our new outpatient letter process.' She attached nothing and gave no other detail. The draft looks professional, but three of its twelve requirements, including a text-message reminder service, were never mentioned by any of the clinic managers she interviewed. What is the most likely cause of the extra requirements?
- AThe model tier she used was too small to keep twelve requirements accurate at once.
- BClaude searched the web and copied requirements from another trust's published process.
- CClaude misread the clinic managers' interview notes and merged their separate wishes.
- DClaude had no stakeholder input to work from, so it filled the gaps with typical requirements. Correct
Why A is wrong: Model choice is a reasonable thing to suspect when output disappoints, but a more capable model given the same empty prompt would still have no interview evidence to work from. The problem is missing input, not reasoning capacity.
Why B is wrong: It is tempting to assume the items came from a real source because they sound specific, but nothing in the situation shows a web search, and Claude does not need one to produce typical requirements. The cause is the absence of her own stakeholder material.
Why C is wrong: Misreading source notes is a real risk in requirements work, which makes this attractive. Here, though, she attached no notes at all, so Claude had nothing to misread.
Why D is correct: With nothing attached, Claude could only draw on what outpatient letter processes commonly include, so it produced plausible requirements such as text reminders that her stakeholders never asked for. Supplying the interview notes and asking Claude to flag gaps as questions would ground the draft.