Intelligence Is Not All You Need: A Case for Artificial Wisdom in Conversational Agents

Abstract

This paper argues that intelligence is the wrong primary design goal for conversational systems. As agents become more capable, proactive, and socially embedded, their most consequential failures stem less from weak task performance than from poor judgment: misframing problems, mishandling uncertainty, overlooking stakeholder interests, and steering users in ways that undermine autonomy and well-being. I propose artificial wisdom as a needed corrective for conversational systems. By artificial wisdom, I mean context-sensitive, morally grounded, meta-cognitively regulated judgment oriented toward human flourishing under uncertainty. This perspective shifts attention from what systems can do to how they should act when advising, persuading, and shaping decisions. I argue that intelligence alone cannot determine appropriate goals, guide action under uncertainty, or ensure beneficial human outcomes. On this basis, I sketch a wisdom-oriented agenda for conversational AI centered on role awareness, meta-cognition, deliberation, self-explanation, and calibrated proactivity. Based in this, I examine its promise, difficulty, and risks.