The Influence of Situational Risk and Caregiver Presence on Expectations for AI Inaction in Smart Eldercare Dilemmas
Description
In smart eldercare settings, AI systems routinely confront ethical dilemmas: when an older adult’s own choices carry risk, should the system intervene to protect safety, or refrain from acting to respect autonomy? Taking the perspective of family caregivers, this research conducted two scenario-based experiments (N = 612) examining how situational risk and caregiver presence relate to expectations for AI inaction. Study 1 (N = 248) found a significant Situational Risk × Caregiver Presence interaction: caregiver presence was associated with markedly higher inaction expectations under low risk, whereas under high risk, inaction expectations were comparably elevated whether or not a caregiver was present. Study 2 (N = 364) reproduced this interaction and, using an exploratory moderated-mediation analysis, found that the association between situational risk and inaction expectations was partially accounted for by two perceptual variables: People’s perceived AI capability and perceived personalized needs of older adults. Caregiver presence moderated both indirect pathways. Because both studies were cross-sectional and vignette-based, these indirect effects are reported as exploratory statistical pathways rather than as tests of causal mechanism, and all conclusions are restricted to what caregivers perceived rather than to what AI systems can objectively do. Implications for the design of AI decision-support in Chinese cultural eldercare are discussed, along with the research designs required to directly test the proposed pathways.
Files
Institutions
- Ningbo UniversityZhejiang, Ningbo