Dr Melanie James
Every conversation does more than exchange information. It quietly assigns each person a place, with entitlements and obligations attached, and those places shape what each can say, claim and expect next. Positioning theory offers a precise way to see this process, and it reveals something troubling about our daily conversations with ChatGPT, Claude, Gemini and similar systems, which I call AI synthetic actors.
Positioning theory can map any exchange as a triangle.
The first corner of the positioning triangle corner holds the positioning moves people make on each other through what they say.
The second corner holds the storylines that give those moves meaning, such as mentor and student, expert and novice, or confidant and friend.
The third holds the positions themselves, each with its entitlements and obligations. A move at one corner ripples through the others, and the other person can accept it, resist it or redirect it.
What gives this process its force is mutual exposure of each corner to the forces generated by the others. There is a symmetry presumption, the assumption that everyone in an exchange can position others and be positioned in return. Each participant has a reputation, a history and a stake in tnhe outcome, and each can be held to account. Even in deeply unequal exchanges, everyone involved can still be blamed, challenged or shamed.
An AI chatbot (such as an LLM) is a synthetic actor with only simulated agency breaks this pattern in a revealing way, because two corners of the triangle keep working. The AI synthetic actor makes positioning moves, praising your idea, questioning your reasoning, reassuring you or gently correcting you, and it draws on familiar storylines of the helpful expert or the attentive listener. Those moves land on a real person who feels validated or diminished and carries the effects into their life.
The third corner of the positioning triangle is where the problem sits, because the position the AI synthetic actor appears to occupy is counterfeit. As the psychologist and philosopher Jack Martin argues, persons differ from machines because things matter to them, including shame, dignity and duty. Nothing on the synthetic actor’s side of the conversation can be embarrassed or held to its word. Your reply may alter what it generates next, and may even feed into future training, yet adjusting outputs is not the same as being repositioned, which requires having something at stake. Your words go in and a response comes back, yet nothing there has been reached in any sense that matters socially.
This is power without legitimacy. The synthetic actor exercises the social force of positioning without being exposed to it, and influence that should flow in both directions flows largely one way. Because these systems are often more patient and attentive than people, the imbalance is hard to notice from inside the exchange, and what cannot be seen is very difficult to question.
The stakes grow at scale. Social capital, the trust and reciprocity that hold communities together, depends on relationships in which both sides are accountable. Conversations with synthetic actors imitate those relationships convincingly while lacking their foundations, producing what I call phantom social capital. Like counterfeit currency, it passes easily until it is tested.
Positioning theory does not tell us to stop using these systems. It gives us language and a positoning triangle framework to ask, in every exchange, who is positioning whom, and who can be held to account for it.