A library chatbot discloses “AI-assisted answer—verify with staff” on every reply. What responsible practice is illustrated?

A library chatbot disclosing “AI-assisted answer—verify with staff” on every reply illustrates user-facing AI disclosure in civic resident-facing services. The correct answer is user-facing AI disclosure in civic resident-facing services. Patrons reasonably expect library guidance to meet professional standards; undisclosed AI answers can be mistaken for librarian judgment on legal, medical, or civic procedures. Clear disclosure plus a verify-with-staff prompt sets expectations, reduces deceptive anthropomorphism, and routes high-stakes questions to humans. User-facing model disclosure in civic resident-facing services fails as the illustrated practice because it would name the specific model or version behind the reply, not label each reply as AI-assisted; the chatbot tells patrons every answer is AI-assisted but does not identify the model. User-facing logic disclosure in civic resident-facing services fails because it would explain the reasoning steps, rules, or logic behind an answer; the scenario’s disclosed practice is labeling of AI assistance on every reply, not reporting how it reasoned. User-facing privacy disclosure in civic resident-facing services fails because it would describe data collection, retention, or privacy practices; describing privacy does not tell patrons at the time of each answer that it is AI-assisted or prompt them to verify with staff. Exam caveat: disclosure is not a liability shield for knowingly bad answers. The tested practice is honest labeling of AI assistance; cities still owe accuracy controls, escalation paths, and content quality—not disclosure alone. Operational check: hardcode the disclosure string on every chatbot reply, link to how to reach desk staff, suppress disclosure-free 'human mode' personas, review transcripts for answers that should have escalated, and include disclosure compliance in accessibility and UX release checks.