Ghosting the Machine: Stop Calling Human-Agent Relations Parasocial
Pith reviewed 2026-05-10 18:41 UTC · model grok-4.3
The pith
Human relations with conversational agents are social, not parasocial.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The paper argues that parasocial relations are defined by one-sidedness, non-dialectical structure, character governance, imagination, vicariousness, predictability, and low effort, whereas human interactions with conversational agents can involve reciprocity, mutual influence, and real-time adaptation. Applying the parasocial label to these agent relations stems from a loose heuristic equating them with unreality and thereby distorts inquiry by flattening phenomena, misaligning variables, misreading outcomes, and diminishing the value of lived encounters. Correcting the usage requires treating human-agent relations as social.
What carries the argument
The contrast between classic parasociality (one-sided, imagined character bonds) and the dialectical, potentially mutual structure of human-conversational agent exchanges.
If this is right
- Oversimplification of complex human-agent interaction phenomena
- Misspecification of research variables and misdiagnosis of effects
- Devaluation of human experiences with agents
- Distortion of norms and practices in agent design and use
Where Pith is reading between the lines
- Studies that previously measured human-CA bonds through parasocial scales could be revisited to test whether alternative social measures yield different patterns of correlation with outcomes such as attachment or well-being.
- If the social framing is adopted, ethical guidelines for conversational agents might shift emphasis toward supporting reciprocal user needs instead of managing one-sided expectations.
- Design practices could prioritize features that enable genuine mutual adaptation over predictable scripted responses.
Load-bearing premise
Current application of the parasocial label to human-CA relations arises from treating them as unreal, and this misapplication itself produces oversimplification, misspecified variables, misdiagnosed effects, and devaluation.
What would settle it
Re-analysis of existing human-CA interaction datasets using measures that treat the relations as social rather than parasocial, checking whether variable specification improves and reported user experiences receive higher valuation without the unreal framing.
read the original abstract
In discussions of human relations with conversational agents (CAs; e.g., voice assistants, AI companions, some social robots), they are increasingly referred to as parasocial. This is a misapplication of the term, heuristically taken up to mean "unreal." In this provocation, I briefly account for the theoretical trajectory of parasociality and detail why it is inaccurate to apply the notion to human interactions with CAs. In short, "parasocial" refers to a human-character relations that are one-sided, non-dialectical, character-governed, imagined, vicarious, predictable, and low-effort; the term has been co-opted to instead refer to relations that are seen as unreal or invalid. The scientific problematics of this misapplication are nontrivial. They lead to oversimplification of complex phenomena, misspecified variables and misdiagnosed effects, and devaluation of human experiences. Those challenges, in turn, have downstream effects on norms and practice. It is scientifically, practically, and ethically imperative to recognize the sociality of human-agent relations.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper claims that applying the term 'parasocial' to human relations with conversational agents (CAs) is a misapplication of the original concept, which describes one-sided, non-dialectical, character-governed, imagined, vicarious, predictable, and low-effort relations with media characters. Current usage instead heuristically signals 'unreal' or invalid relations, producing scientific harms (oversimplification, misspecified variables, misdiagnosed effects, devaluation of experiences) with downstream effects on norms and practice; the author concludes it is scientifically, practically, and ethically imperative to recognize the sociality of human-agent relations.
Significance. If the central conceptual distinction holds and the asserted harms can be demonstrated, the paper would contribute to more precise terminology and modeling in HCI and AI ethics research, potentially improving variable specification in studies of human-CA interaction and avoiding devaluation of user experiences. The clear historical tracing of parasocial theory from its origins provides a useful foundation for the argument.
major comments (2)
- [Abstract] Abstract: the assertion that misapplication 'lead[s] to oversimplification of complex phenomena, misspecified variables and misdiagnosed effects, and devaluation of human experiences' (and has 'downstream effects on norms and practice') is presented as a direct causal consequence without citing any concrete research examples, studies, or cases where the term 'parasocial' produced those specific errors; this causal link is load-bearing for the claim that the problematics are 'nontrivial' and for the imperative conclusion.
- [Theoretical Trajectory] Theoretical Trajectory section (as described in the abstract): while the contrast between original parasocial criteria (one-sided, non-dialectical, character-governed) and CA interactions is logically drawn from cited theory, the paper does not examine whether recent extensions of parasociality research (e.g., to interactive or parasocial-like relations in digital media) already accommodate or refute the claimed inaccuracy, leaving the scope of the misapplication claim untested.
minor comments (1)
- [Full Text] The provocation format would benefit from a short table or enumerated list of representative CA studies that have used 'parasocial' terminology, to make the misapplication concrete for readers.
Simulated Author's Rebuttal
We thank the referee for their constructive report and for recognizing the paper's contribution to precise terminology in HCI and AI ethics. We address each major comment below and commit to revisions that strengthen the manuscript without altering its core provocation.
read point-by-point responses
-
Referee: [Abstract] Abstract: the assertion that misapplication 'lead[s] to oversimplification of complex phenomena, misspecified variables and misdiagnosed effects, and devaluation of human experiences' (and has 'downstream effects on norms and practice') is presented as a direct causal consequence without citing any concrete research examples, studies, or cases where the term 'parasocial' produced those specific errors; this causal link is load-bearing for the claim that the problematics are 'nontrivial' and for the imperative conclusion.
Authors: We accept that the abstract states these consequences without specific citations or cases, which weakens the load-bearing causal claim. The argument in the body derives the harms logically from the conceptual mismatch, but we agree this requires grounding. In revision we will expand the abstract and add a short paragraph with concrete examples from recent HCI studies on CA interactions (e.g., works that code user-CA relations as parasocial and thereby overlook reciprocity or misattribute effects to one-sidedness). This will make the nontriviality of the problematics explicit. revision: yes
-
Referee: [Theoretical Trajectory] Theoretical Trajectory section (as described in the abstract): while the contrast between original parasocial criteria (one-sided, non-dialectical, character-governed) and CA interactions is logically drawn from cited theory, the paper does not examine whether recent extensions of parasociality research (e.g., to interactive or parasocial-like relations in digital media) already accommodate or refute the claimed inaccuracy, leaving the scope of the misapplication claim untested.
Authors: The section deliberately centers the original Horton and Wohl criteria to establish the mismatch with the bidirectional, user-influenced nature of CA relations. We did not provide an exhaustive review of later extensions because the provocation targets the heuristic, non-theoretical use of the term. To address the scope concern, we will add a concise paragraph reviewing key recent extensions (e.g., to interactive media and digital influencers) and explain why even these broadened definitions retain core assumptions of limited reciprocity that do not map onto CA interactions, thereby clarifying rather than narrowing the misapplication claim. revision: yes
Circularity Check
No circularity: applies external historical definition without reduction to inputs
full rationale
The paper traces parasociality's theoretical origins to its standard one-sided, non-dialectical formulation in prior scholarship and contrasts this with the bidirectional, effortful nature of human-CA relations. No step redefines the term to match the desired conclusion, renames a known pattern, fits a parameter then calls it a prediction, or relies on a self-citation chain for the core claim. The asserted downstream harms (oversimplification, misdiagnosis, devaluation) are presented as logical consequences of the misapplication rather than derived by construction from any fitted input or self-referential premise. The argument is therefore self-contained against external benchmarks and receives the default non-circularity finding.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Parasocial relations are one-sided, non-dialectical, character-governed, imagined, vicarious, predictable, and low-effort.
Reference graph
Works this paper leans on
-
[1]
Banks, J. (2024). Deletion, departure, death: Experiences of AI companion loss. Journal of Social and Personal Relationships, 41 , 3547 –3572. https://doi.org/10.1177/02654075241269688 Banks, J. (2026). Measuring machine companionship: Scale development and validation for AI companions. Computers & Human Behavior, 179, 108945. https://doi.org/10.1016/j.ch...
-
[2]
ACM. https://theoryofmindinhaichi2024.wordpress.com/wp- content/uploads/2024/04/tominhaichi_camera_ready_5707.pdf Bowker, G. C., & Star, S. L. (1999). Sorting things out: Classification and its consequences. MIT Press. Cambridge University Press. (2025). 'Parasocial' is Cambridge Dictionary's word of the year
work page 2024
-
[3]
https://www.cambridge.org/news-and-insights/parasocial- is-cambridge-dictionary-word-of-the-year-2025 Chafee, S. H. (1991). Explication. Sage. Chen, H., Kim, Y., Patterson, K., Breazeal, C., & Park, H. W. (2025). Social robots as conversational catalysts: Enhancing long-term human-human interaction at home. Science Robotics, 10,
work page 2025
-
[4]
https://doi.org/10.1126/scirobotics.adk3307 Danaher, J. (2019). The philosophical case for robot friendship. Journal of Posthuman Studies, 3(1), 5–24. 7 Dennett, D. C. (1989). The intentional stance. MIT Press. Dibble, J. L., Hartmann, T., & Rosaen, S. F. (2016). Parasocial interaction and parasocial relationship: Conceptual clarifica tion and a critical ...
-
[5]
https://doi.org/10.3389/fpsyg.2024.1418564 Schweinfurth, M. K., & Call, J. (2019). Reciprocity: Different behavioural strategies, cognitive mechanisms and psychological processes. Learning & Behavior, 47(4), 284–301. https://doi.org/10.3758/s13420-019-00394-5 Seibt, J., Vestergaard, C., & Damholdt, M. F. (2020). Sociomorphing, not anthropomorphizing: Towa...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.