Study: AI Companions Could Deepen Social Inequality

New research from Singapore Management University (SMU) and Duke-NUS Medical School offers a critical examination of the increasing adoption of AI companions. The study argues that the pertinent question is not whether these technologies are beneficial or harmful, but rather who benefits, who is most at risk, and how to ensure equitable outcomes.

Published in Nature Human Behaviour, the research introduces a novel framework for understanding how AI companions might exacerbate existing social inequalities. Rather than viewing AI companions through a simple dichotomy of good or bad, the researchers analyzed how disparities in user social support, technology design, and governance can lead to unequal results. They also propose practical recommendations for policymakers, developers, and users.

The study highlights a "rich-get-richer" dynamic in human relationships, where individuals with strong social networks may use AI companions to supplement existing connections. Conversely, those who are lonely or socially isolated may rely on AI companions as a substitute for human interaction, potentially leading to a gradual erosion of social skills over time.

The research adapted a framework based on the "Swiss cheese model" to explain risk emergence: failures in multiple protective layers, including user AI literacy, social support, platform design, and regulation, can leave vulnerable users more exposed to harm. Governance was identified as the most significant gap among the four layers examined, as AI companions remain in a regulatory grey area in many countries.

Researchers recommend that AI companions be treated as health-related technologies, enabling stronger safeguards such as age-appropriate design requirements, transparency about interacting with AI, limitations on manipulative features, and clearer data governance and crisis response standards. This is particularly relevant in Singapore, given its widespread technology adoption.