Beyond Comfort: Interesting Experiences and the Measurement of Well-Being in AI Care Companions

Abstract

AI care companions for older adults are typically evaluated by whether they reduce loneliness, improve mood, provide companionship, or keep users mentally active. These are genuine benefits, but they do not exhaust what matters to well-being. Drawing on Lorraine Besser’s account of interesting experiences, this paper develops two reasons to treat robust cognitive engagement as a distinct design and evaluation target. First, if interestingness has experiential value not reducible to pleasure, comfort-centered measures omit an important prudential good. Second, even if interestingness is understood as a cognitively structured form of pleasure, it has a prolific character; every genuinely interesting experience opens or sustains some further possibility for engagement, although the extent of this opening varies and its development into later inquiry, activity, or relationship is contingent. Through a philosophical case study of the AI-powered care companion ElliQ, we show how a welfare-oriented companion can provide genuine benefits while failing to identify or measure this dimension of experience. Because personalized and proactive companions can shape—and sometimes displace—users’ finite time and attention, we argue that they bear a conditional, role-based, pro tanto responsibility not to narrow users’ experiential worlds. We translate this responsibility into three design principles: situated novelty, calibrated challenge, and continuity and outward transfer. These principles aim to preserve voluntary, permission-sensitive opportunities for curiosity, reflection, and engagement beyond the system itself. AI care companions should be judged not only by whether they make a person’s life more comfortable, but also by whether they help keep their world open.