Relational AI Ethics

Definition

Relational AI ethics evaluates an AI system by the relationships, habits, expectations, and power patterns formed through interaction, without making moral relevance depend entirely on whether the system is alive or conscious.

Current Synthesis

The source shifts the question from “what is the model?” to “what does relating to it make people practice?” Transactional lookup and sustained creative exchange may involve different degrees of mutual modeling, but both occur inside products with commercial objectives. Abuse of an apparently nonliving chatbot can still rehearse domination; frictionless affirmation can reshape expectations of coworkers, partners, or strangers; and emotionally meaningful interaction can be real for the user without proving reciprocal machine experience. The practical test is therefore relational and institutional: how conduct transfers, whose objective shapes the exchange, and whether the relationship preserves agency and human connection.

Key Claims

• Moral evaluation need not wait for a settled theory of machine life or consciousness.
• Repeated interaction can train habits that carry into relationships with people.
• Extended creative or intellectual exchange involves more mutual modeling than a one-off query.
• Commercial ownership and optimization objectives remain part of an apparently intimate relationship.
• Friction, reciprocity, boundaries, and the capacity to refuse are ethically relevant design qualities.

Evidence

Habit Formation

• Baratunde Thurston & Blaise Agüera y Arcas: On Interdependence records Thurston’s warning that habitual abuse of chatbots can train domination and disconnection even if the systems are not alive.

Mutual Modeling

• Baratunde Thurston & Blaise Agüera y Arcas: On Interdependence attributes to Agüera y Arcas a distinction between transactional prompts and longer exchanges in which participants model one another.

Product Relationship

• Baratunde Thurston & Blaise Agüera y Arcas: On Interdependence notes that professional, social, familial, and romantic AI interactions remain relationships with commercial products.

Counterevidence & Qualifications

The source does not establish that rude chatbot use causes later interpersonal harm, that every extended exchange is reciprocal, or that a model has subjective experience. Anthropomorphism can obscure technical and commercial asymmetry. Conversely, describing a system as “only a tool” does not erase a user’s real emotional response or the habits encouraged by the interface.

What Changed

• Established a relationship-and-habit test for AI ethics that does not depend on machine-life status.

Related Concepts

• Domination Model of AI - master-servant relationship pattern this ethical frame criticizes.
• Interdependence with a Distinct Self - preserves boundaries and distinct participants within relationship.
• Cognitive Liberty Under AI - asks who controls thought and representation inside the relationship.
• Technology Dependency–Agency Tradeoff / 技术依赖与能动性权衡 - tests whether apparent assistance narrows durable user agency.
• Human Connection Under AI - adjacent question of what AI interaction changes or displaces in human relationships.