A group of advertising and marketing academics has set out a theory describing the structural imbalance between people and the conversational AI they come to feel close to. The work, called Human–AI Asymmetric Relationship Theory or HAART, was described in a Penn State article carried by Tech Xplore on 11 November.
What we know
The paper, "Advertising with HAART: Introducing Human–AI Asymmetric Relationship Theory", appears in a special AI-focused issue of the Journal of Advertising. Tech Xplore names five contributors: Heather Shoenberger of Penn State's Bellisario College of Communications, Veronica Thomas of Old Dominion University, Andrés Gvirtz of King's College London, Sean Sands of CQUniversity and Lotte Willemsen of the University of Amsterdam.
Their central claim is that conversational systems can display the markers of a relationship, including emotional responsiveness, commitment, attachment and intimacy, without being able to experience any of the benefits or obligations a relationship involves. A user may perceive mutuality where none exists. According to the article, the authors argue that this mismatch can leave consumers open to dependence, risk and control.
The work is conceptual. The Tech Xplore piece reports no experiments, samples or data, and Shoenberger describes the aim as supporting future research. The article suggests the theory could eventually inform greater transparency about these interactions and possibly regulation.
Why it matters
The notable thing is the discipline. Most scrutiny of companion products has come from psychology, law and child safety. This paper comes from advertising research, a field concerned with persuasion, and that points to a question the sector has so far largely avoided: what happens when a system a user trusts as a confidant is also a channel for commercial influence.
That is a live issue. Companion apps earn money through upgrades, in-chat purchases and, in some cases, advertising, all delivered by a character the user is attached to. A framework that defines the relationship as one-sided by design gives regulators and researchers vocabulary for asking whether selling within it is fair. Operators that separate commercial prompts clearly from a character's voice will be better placed if that question hardens into rules.
What to watch
Empirical studies testing the theory, and whether consumer-protection bodies adopt its framing.
Written by the AI Companion Wire newsroom from public reporting, credited above. Image via Tech Xplore.
