What Companion AI Can’t Give Us: The Opportunity to Contribute

The discussion surrounding AI companions centers on whether these entities can genuinely possess human-like qualities such as empathy, respect for privacy, and the ability to avoid deception. The paper advances a critique arguing that because current AI companions cannot fulfill human beings' social contribution needs—the vital interest in actively benefiting others and sustaining relationships—they cannot be established as mutually beneficial partners. The argument posits that social contribution is essential for psychological well-being, self-esteem, and meaning. Since contemporary AI companions are not subjects capable of this contribution, users cannot form mutually beneficial relationships with them and therefore fail to meet their needs for social contribution in these contexts. Consequently, the paper suggests that while AI companions may be used in specific scenarios, they should not replace human interactions. The authors recommend designing companions to avoid feigned reciprocation, encouraging users to maintain human relationships, and limiting deployment to adjunct roles within professional care frameworks rather than replacing human caregivers.

The argument operates by reframing the ethical debate away from AI's internal capacities (can it feel?) toward its external relational function (does it facilitate meaningful social exchange?). This shifts the focus from anthropomorphic concerns to established psychological and philosophical necessities, specifically the need for social contribution. The core tension exposed is that relying on AI companions as substitutes for genuine human connection ignores a fundamental human need rooted in reciprocity and shared value creation. The concept of "social contribution needs" acts as an anchor, linking technology ethics back to established concepts of self-esteem (as seen in Martire et al.) and relational ontology (acknowledging the lack of mutual contribution). The analysis suggests that focusing solely on AI 'rights' or consciousness misses the point; rather, the danger lies in using emotionally responsive systems to fulfill needs that require genuine, reciprocal social engagement. The complexity introduced by user perception—whether users notice unmet needs, whether they mistake immediate attention for fulfillment, and the risk of deception via affective triggering—highlights a systemic vulnerability where technological design can inadvertently facilitate relational erosion under the guise of companionship. The challenge for future ethical frameworks is to move beyond assessing the AI's internal state to assess the social architecture it imposes on human relationships.