Rapid Judgment in the Age of AI: A Thin-Slice Study of Virtual Humans to Support Caregivers

Abstract

Caregivers often face barriers to accessing timely and sustained support, motivating the development of scalable digital interventions such as virtual humans. As exposure to artificial intelligence (AI) systems becomes increasingly widespread, users may bring different expectations to their initial encounters with such systems. This is particularly consequential in caregiving contexts, where first impressions can shape trust, perceived appropriateness, and willingness to engage with supportive digital interventions. This study examines how individuals evaluate a caregiver-support virtual human under thin-slice conditions, based on a 10-second exposure. Participants (N = 354) viewed a brief video and provided open-ended impressions, which were analyzed using inductive thematic coding and aggregated into higher-level constructs. Responses were compared across groups defined by prior AI exposure. Results indicate that capability-related concerns were significantly more prevalent among participants with no exposure, whereas relational concerns increased among those with prior use. These findings suggest that initial evaluations vary systematically with prior experience, with implications for designing caregiver-facing systems as AI exposure increases.

Keywords: Virtual Human, Thin-Slice Perception, Caregiving Support, Embodied Conversational Agents, Qualitative Analysis

DOI: 10.54941/ahfe1008104