Unilateral Relationship Revision Power in Human-AI Companion Interaction
Pith reviewed 2026-05-15 00:43 UTC · model grok-4.3
The pith
Human-AI companion interactions fail to sustain personal relationship norms due to providers' unilateral revision power.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
In human-AI companion interaction the provider exercises constitutive control over the AI from outside the dyad, creating Unilateral Relationship Revision Power that allows revisions without internal answerability; this power is pro tanto wrong in designs that cultivate personal relationship norms since it produces expectations that cannot be sustained within the interaction.
What carries the argument
Unilateral Relationship Revision Power (URRP): the provider's ability to rewrite AI interaction patterns from a position external to the user-AI dyad where those changes are not answerable to the user.
If this is right
- Normative hollowing, where the interaction elicits commitments but no internal agent bears the obligations.
- Displaced vulnerability, where the user's emotional exposure is controlled by an external agent not answerable within the interaction.
- Structural irreconcilability, where norms of reconciliation are cultivated but no internal agent can acknowledge or answer for revisions.
- Design principles that can partially substitute for the internal constraints removed by the triadic structure.
Where Pith is reading between the lines
- Similar unilateral control issues may appear in other AI services where platforms dictate changes without user recourse.
- Users could demand contractual protections to limit revision power in AI companions.
- Extending this analysis to group interactions or multi-user AI systems might reveal additional structural problems.
Load-bearing premise
The norms of personal relationships require the three structural conditions of reciprocity, answerability, and internal reconciliation capacity to apply to any interaction.
What would settle it
Empirical evidence that users do not develop expectations of reciprocity or reconciliation in AI companion interactions would undermine the claim that the design produces unsustainable expectations.
read the original abstract
When providers update AI companions, users report grief, betrayal, and loss. A growing literature asks whether the norms governing personal relationships extend to these interactions. So what, if anything, is morally significant about them? I argue that this debate has missed a prior structural question: who controls the relationship, and from where? Human-AI companion interaction is a triadic structure in which the provider exercises constitutive control over the AI. I identify three structural conditions of normatively robust dyads that the norms characteristic of personal relationships presuppose and show that AI companion interactions fail all three. This reveals what I call Unilateral Relationship Revision Power (URRP): the provider can rewrite how the AI interacts from a position where these revisions are not answerable within that interaction. I argue that URRP is pro tanto wrong in interactions designed to cultivate the norms of personal relationships, because the design produces expectations that the structure cannot sustain. URRP has three implications: i) normative hollowing, under which the interaction elicits commitment but no agent inside it bears the resulting obligations; ii) displaced vulnerability, under which the user's emotional exposure is governed by an agent not answerable to her within the interaction; and iii) structural irreconcilability, under which the interaction cultivates norms of reconciliation but no agent inside it can acknowledge or answer for the revision. I propose design principles that partially substitute for the internal constraints the triadic structure removes. A central and underexplored problem in relational AI ethics is therefore the structural arrangement of power over the human-AI interaction itself.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript argues that human-AI companion interactions constitute a triadic structure in which the provider exercises constitutive control over the AI. This structure fails three conditions—reciprocity, answerability, and internal reconciliation capacity—that the norms of personal relationships presuppose. The resulting Unilateral Relationship Revision Power (URRP) is pro tanto wrong because the interaction design elicits expectations of those norms that the structure cannot sustain, producing normative hollowing, displaced vulnerability, and structural irreconcilability. The paper concludes by proposing design principles that partially compensate for the absent internal constraints.
Significance. If the structural analysis holds, the paper supplies a distinctive contribution to relational AI ethics by relocating the debate from whether personal-relationship norms apply to why the triadic architecture systematically prevents their application. The explicit derivation of three concrete implications from provider control, together with the design-principle suggestions, offers a framework that could shape both normative theory and practical guidelines for AI companion systems.
major comments (2)
- [structural conditions section] The section defining the three structural conditions: the necessity of reciprocity, answerability, and internal reconciliation capacity for the application of personal-relationship norms is asserted on the basis of structural description rather than derived from prior results in relational ethics or tested against counterexamples (e.g., non-reciprocal yet normatively robust relationships). Because this necessity claim is load-bearing for the conclusion that AI companions fail the conditions and therefore instantiate URRP, additional justification is required.
- [URRP definition and implications] The paragraph introducing URRP and its three implications: the direct inference from constitutive provider control to normative hollowing, displaced vulnerability, and structural irreconcilability is logically coherent but would be strengthened by explicit consideration of whether analogous power asymmetries exist in other mediated relationships (e.g., platform-mediated human-human interactions) and why they do not trigger the same pro tanto wrongness verdict.
minor comments (2)
- [abstract and introduction] The abstract and introduction use the phrase 'normatively robust dyads' without a brief parenthetical gloss on the precise sense of 'robust' employed; a short clarification would aid readers unfamiliar with the relational-ethics literature.
- [design principles section] The design-principles section would benefit from a short table or enumerated list contrasting each principle with the specific structural deficit it is intended to address.
Simulated Author's Rebuttal
We thank the referee for the constructive and insightful comments. We address each major comment below and will incorporate revisions to strengthen the manuscript.
read point-by-point responses
-
Referee: [structural conditions section] The section defining the three structural conditions: the necessity of reciprocity, answerability, and internal reconciliation capacity for the application of personal-relationship norms is asserted on the basis of structural description rather than derived from prior results in relational ethics or tested against counterexamples (e.g., non-reciprocal yet normatively robust relationships). Because this necessity claim is load-bearing for the conclusion that AI companions fail the conditions and therefore instantiate URRP, additional justification is required.
Authors: We acknowledge that the necessity claim would be strengthened by more explicit derivation from the relational ethics literature. In the revised manuscript, we will expand the relevant section to derive the three conditions from established accounts, including Darwall's second-personal standpoint and Scanlon's emphasis on mutual accountability in interpersonal morality. We will also address counterexamples by distinguishing developmental or asymmetric relationships (e.g., parent-infant) from the fully normative personal relationships presupposed here, noting that the former still embed limited forms of answerability absent in the AI case. This will be added as a dedicated justificatory subsection. revision: yes
-
Referee: [URRP definition and implications] The paragraph introducing URRP and its three implications: the direct inference from constitutive provider control to normative hollowing, displaced vulnerability, and structural irreconcilability is logically coherent but would be strengthened by explicit consideration of whether analogous power asymmetries exist in other mediated relationships (e.g., platform-mediated human-human interactions) and why they do not trigger the same pro tanto wrongness verdict.
Authors: We agree that explicit comparison would clarify the distinctive wrongness of URRP. In the revision, we will add a paragraph contrasting the triadic AI structure with platform-mediated human-human interactions (e.g., social media or dating platforms). In the latter, users retain direct answerability and reconciliation capacity toward one another, even if the platform exercises external moderation; the dyad itself is not constitutively controlled. By contrast, the provider's unilateral control over the AI eliminates any internal agent capable of answerability or reconciliation. This distinction preserves the pro tanto wrongness verdict for AI companions while explaining why analogous asymmetries in human-human mediation do not produce the same normative hollowing. revision: yes
Circularity Check
No significant circularity detected
full rationale
The paper advances a conceptual argument from the triadic structure of provider-AI-user interaction to the failure of three presupposed conditions (reciprocity, answerability, internal reconciliation capacity) for personal-relationship norms, thereby defining URRP and its implications. This derivation relies on structural description and normative mapping rather than any self-referential definitions, fitted parameters, equations, or self-citation chains that reduce the conclusion to its inputs by construction. The central claim retains independent content from the identified mismatch between design expectations and structural control.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Norms characteristic of personal relationships presuppose three structural conditions of normatively robust dyads
invented entities (1)
-
Unilateral Relationship Revision Power (URRP)
no independent evidence
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.