Litware is designing a multi-agent workplace assistant that can draft emails, propose calendar changes, and call internal HR tools. Legal and UX require transparency, user control, and minimization of harmful or over-autonomous behavior. Which design package best supports human-AI experience (HAX) and Responsible AI principles?
Select an answer to reveal the explanation.
Short Explanation
A is the HAX + Responsible AI package the exam wants. Users should understand what agents are doing, preview consequential actions, confirm state changes, undo or escalate, and trust that sensitive HR tools are constrained with refuse/handoff rules—and that actions are auditable. B optimizes for demos, not safety. C is deceptive and noncompliant. D trades away transparency. Multi-agent UX is part of architecture: autonomy levels, confirmation patterns, and behavioral guidelines are first-class design artifacts, not afterthoughts.
Full Explanation
A is correct because multi-agent logical architecture must include controls supporting human-AI experience and Responsible AI: transparency, appropriate autonomy, human oversight, and constrained tools. Action previews, confirmations, undo/escalation, audit logs, and refuse guidelines operationalize those principles. B is incorrect: opaque auto-execution increases harmful autonomy and reduces user trust and control. C is incorrect: undisclosed executive impersonation violates transparency and organizational policy norms. D is incorrect: removing explanations undermines informed user oversight. Pair HAX patterns with technical enforcement (tool gates), not UI copy alone.