A recent study by Singapore Management University (SMU) and Duke-NUS Medical School has introduced a new framework for understanding how AI companions may exacerbate social inequality. Published in Nature Human Behaviour, the research argues that the focus should shift from debating whether AI companions are beneficial or harmful to examining who benefits, who is at risk, and how to ensure equitable use of AI technology.
The study, led by Zhang Qiyang from SMU and Liu Nan from Duke-NUS, highlights the potential for AI companions to reinforce existing inequalities. It identifies a “rich-get-richer” dynamic, where individuals with strong social networks use AI to enhance relationships, whilst those who are isolated may rely on AI as a substitute for human interaction, risking social deskilling.
The researchers adapted the Swiss cheese model to illustrate how multiple small failures in user literacy, social support, platform design, and governance can align to create systemic risks. Governance emerged as a critical gap, with AI companions often regulated as consumer applications rather than health-related technologies.
The study suggests recognising AI companions as technologies with psychosocial implications, advocating for age-appropriate design, transparent AI interaction disclosures, and robust privacy standards. It also recommends a layered governance approach, combining regulation with efforts to improve AI literacy and encourage responsible platform design.
Singapore is highlighted as both well-positioned and vulnerable to the rapid adoption of AI companions, given its digital infrastructure and societal conditions. The study calls for proactive measures to ensure AI supports human wellbeing without replacing meaningful connections.
