AI Companion Robot Data Sharing: Preferences of an Online Cohort and Policy Implications

Abstract

AbstractPolicymakers have recognized the urgent need to create AI data protections, yet the interests of older adults have thus far not been well represented. We report peoples' perspectives on small AI companion robots for older adults, along with attendant issues related to facial expression and conversation data collection and sharing. Data are from a cross‐sectional survey of an online cohort of the Oregon Center for Aging & Technology at Oregon Health & Science University, with a response rate of 45% and analytic sample of 825 (mean age: 63.9, rang: 25‐88). Logistic regressions examined relationships between comfort and data sharing preferences with socio‐demographic characteristics. Just over half (52.3%) were somewhat or very comfortable with an artificial companion robot during the pandemic and 45.2% were under normal circumstances. In adjusted models, being younger, male, and having lower formal education and greater confidence in computer use were associated with greater likelihood of being comfortable with a companion robot. Those who were at least somewhat comfortable with robots recording their conversations (15%) or reported that they would probably want their facial expressions read for emotion detection (52.8%) also selected with whom they want these data shared. Free text comments were thematically analyzed. Primary themes were that robot‐based data collection constitutes over monitoring and invasion of privacy, with participants predicting data privacy, security, and use issues. These findings about the importance potential users place on data protection and transparency demonstrate a need for law and policy to act to enable trustworthy, desirable companion robots.