A housing authority assistant answers eligibility questions in a tone that sounds like a final binding determination. What GenAI output risk should the platform owner address first?
Select an answer to reveal the explanation.
Short Explanation
If the bot sounds like a judge handing down an eligibility verdict, residents will treat soft guidance like hard law. That authority-inflation risk needs decision-boundary limits and clearer 'not a final determination' framing. Logo compression and SPF records are not the issue here.
Full Explanation
Authority inflation occurs when generative answers use tone, certainty, or wording that implies binding decisions the system is not empowered to make. A housing authority assistant that answers eligibility questions like a final determination can lead residents to treat guidance as an official ruling, creating fairness, appeals, and legal-exposure risk. Platform owners should address decision-boundary language first: clarify advisory status, route determinations to authorized systems and staff, and prevent the model from sounding like a judge.
Excessive JPEG compression on logo assets is an image-quality issue with no bearing on decisional tone in eligibility answers. Missing SPF records on a marketing subdomain affect email authentication for that domain; they do not cause a chatbot to imply binding eligibility outcomes. Underutilized CDN cache for static policy PDFs is a performance/cost concern, not the GenAI output risk of authority inflation.
Exam caveat: even accurate policy quotes can inflate authority if framed as "you are approved" versus "here is how the published policy reads; a caseworker must decide." Hallucination and authority inflation can co-occur but are distinct—correct policy text delivered as a verdict is still inflation.
Operational check: enforce system prompts and UI disclaimers that answers are informational; block disposition verbs for eligibility; escalate determination requests to human workflows; test red-team prompts that demand final rulings; and align legal counsel on allowed language. Housing platforms should log when residents ask for decisions and ensure the bot never substitutes for an authorized determination letter.