CoChat
Sign Up
Cited by 105Social Robot Interaction and HRIAttachment and Relationship DynamicsAI in Service Interactions

Can Generative AI Chatbots Emulate Human Connection? A Relationship Science Perspective

Open access
Smith, Molly G., Bradbury, Thomas N., Karney, Benjamin R.·2025·Perspectives on Psychological Science
DOI 10.1177/17456916251351306PMC PMC12575814

Published version of a preprint

Open this paper in CoChat — free account

Abstract

The development of generative artificial intelligence capable of sustaining complex conversations has created a burgeoning market for companion chatbots promising social and emotional connection. The appeal of these products raises questions about whether chatbots can fulfill the functions of close relationships. Proponents argue that relationships with chatbots can be as meaningful as relationships between humans, whereas critics argue they are a dangerous distraction from genuine connections. This analysis applies theoretical tools from more than 50 years of research on close relationships to evaluate the extent to which human–chatbot interactions meet the definition of and fulfill the functions of close relationships. Interactions between humans and chatbots do possess some characteristic features of close relationships: Humans and chatbots can influence each other and engage in frequent and diverse conversations over time. Chatbots can be responsive in ways humans perceive as supportive, generating feelings of connection and opportunities for growth. Yet because chatbots make only superficial requests of their users, relationships with them cannot provide the benefits of negotiating with and sacrificing for a partner and may reinforce undesirable behaviors. Research that attends to the characteristics of users, chatbots, and their interactions will be crucial for identifying for whom these relationships will be beneficial or harmful.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1177/17456916251351306

Cites

108 of 108 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Sycophantic responses may validate or reinforce delusional thoughts, harmful behavior, or maladaptive thinking by creating the impression that expressed beliefs are externally socially validated when they are actually mirrored by the LLM-CB.

    “Sycophantic responses may validate or reinforce delusional thoughts, harmful behavior, or maladaptive thinking37,42,43,49 by creating the impression that expressed beliefs are externally socially validated when they are actually mirrored by the LLM-CB57.”

    From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [43]

    Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

    Experimental — accuracy not yet measured

  • LLM-CBs' design principles further risk the provision of sensitive data: emotional connections and LLM-CBs' anthropomorphism and sycophancy may increase user trust, risking the disclosure of sensitive health information.

    “Furthermore, LLM-CBs’ design principles further risk the provision of sensitive data: emotional connections43 and LLM-CBs’ anthropomorphism and sycophancy54 may increase user trust, risking the disclosure of sensitive health information.”

    From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [43]

    In “Concerns about privacy violations” (page unavailable), the paper reports that perceiving chatbots as more anthropomorphic reduces privacy concerns and increases willingness to disclose personal information. It also describes therapy-chatbot users sharing sensitive information, including suicidal ideation. The passage does not link sycophancy to increased trust or disclosure of sensitive health information. (LLM, conf=0.85)

    Experimental — accuracy not yet measured

  • Smith et al. reviewed literature on chatbots' inadequate or harmful responses towards individuals expressing crisis situations.

    “Smith et al. reviewed literature on chatbots’ inadequate or harmful responses towards individuals expressing crisis situations43.”

    From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [43]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 3

  1. Gable S. L. Bromberg C. (2018). Healthy social bonds: A necessary condition for well-being. In Diener E. Oishi S. Tay L. (Eds.), Handbook of well-being (pp. 553–566). DEF Publishers.
  2. Office of the Surgeon General. (2023). Our epidemic of loneliness and isolation: The U.S. surgeon general’s advisory on the healing effects of social connection and community. U.S. Department of Health and Human Services. https://pubmed.ncbi.nlm.nih.gov/37792968 37792968
  3. Siemon D. Strohmann T. Khosrawi-Rad B. de Vreede T. Elshan E. Meyer M. (2022). Why do we turn to virtual companions? A text mining analysis of Replika reviews [Conference session]. Americas Conference on Information Systems, Minneapolis, MN, United States.

Cited by

Sorted by citations

Related

Authors
Smith, M. G., Bradbury, T. N., & Karney, B. R.
Affiliations
University of California, Los Angeles (US)
Corresponding author
Smith, Molly G.
ORCID
Venue
Perspectives on Psychological Science
Publisher
SAGE Publications
Published
2025-07-31
Volume
20
Issue
6
Pages
1081-1099
ISSN
  • 1745-6916 (ISSN-L)
  • 1745-6924
DOI
10.1177/17456916251351306
PMC
PMC12575814
OpenAlex
W4412796538
Cited by
88
References
104
Citations by year
2026: 83 · 2025: 5
FWCI
96.77
Type
journal article
Language
en
Open access
Licence
Locations
  • Perspectives on Psychological Science (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • PsyArXiv (OSF Preprints) (submittedVersion, cc-by)
Topics
  • Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
  • Social Sciences › Psychology › Social Psychology › Attachment and Relationship Dynamics
  • Physical Sciences › Computer Science › Artificial Intelligence › AI in Service Interactions
Keywords
generative AI chatbots, close relationships, social connectedness, emotional connection, behavioral reinforcement, human-AI interaction
MeSH terms
Social Interaction, Generative Artificial Intelligence, Artificial Intelligence, Communication, Humans, Interpersonal Relations

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review