A public-sector multi-agent assistant will recommend benefit options. Leadership requires human-AI experience (HAX) and responsible AI controls in the logical design—not only accuracy metrics. Which design package best addresses this?
Select an answer to reveal the explanation.
Short Explanation
You need to say you’re AI, let people edit or reject recommendations, hand off to humans, ground policy claims, and keep an audit trail. That’s the human-AI experience and responsible AI control set the exam tests. Dark patterns like fake human impersonation are the opposite.
Full Explanation
The correct design package provides clear AI disclosure, user control to edit or reject recommendations, escalation paths to human case workers, grounded citations where policies are cited, and logging for accountability. These controls support human-AI experience by preserving transparency, user agency, human oversight, contestability, grounding, and accountability.
The package that hides AI outputs, auto-submits applications without review, and optimizes only for conversion violates transparency and human oversight.
The package that impersonates a named case worker is deceptive and unauthorized.
The package that removes uncertainty language overclaims certainty and can create harmful overreliance.
Exam tip: Responsible AI multi-agent answers emphasize transparency, human oversight, contestability, and grounding—not conversion hacks.