A county poll-book checkout script from an LLM looks tidy but calls screens and APIs the election app does not expose, so it will not run. Which GenAI defect does this represent?
Select an answer to reveal the explanation.
Short Explanation
A script can look polished and still be fiction. Calling screens and APIs that are not there is a classic hallucination: incorrect or non-functioning testware.
Full Explanation
A county poll-book checkout script from an LLM that looks tidy but calls screens and APIs the election app does not expose—so it will not run—is a hallucination that generates incorrect or non-functioning test scripts. Civic election systems require runnable, interface-true artefacts; polished fiction is still defective testware. Bias that omits accessibility checks fails as the wrong class for non-existent controls. Non-determinism because two runs differed fails when the stem emphasizes impossible UI/API calls, not run-to-run variance. A reasoning error limited to fee-ladder math only fails by mismatch to election checkout interfaces. Exam caveat: non-functioning scripts that invent controls are hallucinations. Operational check: execute or statically validate civic poll-book scripts against the real app map, quarantine calls to missing screens/APIs, regenerate with an interface inventory attached, and block promotion until scripts start cleanly in the election test lab. Retain the municipal audit note so later testers can see why the decision protects public service quality and fairness. Retain the municipal audit note so later testers can see why the decision protects public service quality and fairness.