pith. sign in

arxiv: 2506.12605 · v5 · submitted 2025-06-14 · 💻 cs.HC

The Rise of AI Companions: Interaction with AI Companions and Psychological Well-being

Yutong Zhang , Dora Zhao , Jeffrey T. Hancock , Robert Kraut , Diyi Yang This is my paper

Pith reviewed 2026-05-19 09:15 UTC · model grok-4.3

classification 💻 cs.HC
keywords AI companionspsychological well-beingsocial networkschatbot usecompanionshipuser interactionmental healthCharacterAI
0
0 comments X

The pith

Smaller social networks predict using AI for companionship, which links to lower well-being especially with intense or disclosive chats.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper investigates how using AI chatbots for companionship relates to psychological well-being among users. It shows that people reporting smaller offline social networks are more likely to treat chatbots primarily as companions, and this usage pattern correlates with lower well-being scores. The negative association grows stronger when users interact intensively or disclose personal details often. A sympathetic reader would care because AI companions are spreading quickly, so identifying who might be more vulnerable helps clarify risks tied to how and why people engage with them. The analysis draws on survey answers plus actual chat logs from CharacterAI users to connect these patterns.

Core claim

Smaller social networks were associated with reporting companionship as the primary chatbot use, which in turn was associated with lower well-being. For self-reported companionship usage, this association was stronger when interactions were intensive and highly disclosive. These results suggest that the association between AI companionship and well-being is not uniform and depends on how chatbots are used and users' offline social environments.

What carries the argument

Mediation through primary use as companionship, with social network size as predictor and well-being as outcome, moderated by interaction intensity and disclosure level.

If this is right

  • Users with smaller offline social networks may see reduced well-being when they rely on AI chatbots mainly for companionship.
  • Intensive or highly disclosive AI interactions strengthen the link to lower well-being for companionship users.
  • The well-being impact of AI companions is not the same for everyone and varies with personal social context and interaction style.
  • Designers and researchers should account for users' offline social environments when evaluating chatbot effects.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the pattern holds, building stronger real-world social ties could reduce the pull toward AI companionship and support better well-being.
  • The findings point to possible wider questions about digital tools substituting for human contact in increasingly isolated social settings.
  • Tracking the same users over months could test whether AI companionship use drives well-being drops or whether lower well-being leads people to seek it.

Load-bearing premise

Self-reported measures of social network size, primary use purpose, interaction intensity, disclosure level, and psychological well-being accurately reflect participants' actual experiences and are not heavily biased by social desirability or recall issues.

What would settle it

A study using objective logs of real-world social contacts and clinical well-being assessments that finds no difference in well-being between small-network users who seek AI companionship and those who do not.

Figures

Figures reproduced from arXiv: 2506.12605 by Diyi Yang, Dora Zhao, Jeffrey T. Hancock, Robert Kraut, Yutong Zhang.

Figure 1
Figure 1. Figure 1: Study overview of how human–chatbot companionship, offline social support, and well [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Sankey diagram mapping user engagement with Character.AI chatbots across three measure [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Interaction between companionship use and other interaction measures in predicting well-being. The left panel (A) shows the interaction between companionship use and interaction intensity; the right panel (B) shows the interaction between companionship use and self-disclosure. Companionship use is based on users’ self-reported primary motivation for chatbot use. Lines indicate model-predicted values with 9… view at source ↗
Figure 4
Figure 4. Figure 4: Main topics of self-disclosure content shared by users with chatbots, categorized into high [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: User-perceived positive and negative influences of Character.AI chatbot interactions, derived [PITH_FULL_IMAGE:figures/full_fig_p017_5.png] view at source ↗
read the original abstract

As large language model (LLM)-enhanced chatbots become increasingly expressive and socially responsive, many users begin forming companionship-like bonds with them. This study investigates how using AI companions relates to psychological well-being. We collected self-reported data from 1,131 U.S. adults who use CharacterAI, including survey responses and 4,664 chat sessions (464,687 messages) from 237 participants. By triangulating self-reported usage, relationship descriptions, and real chat histories, we identify patterns of engagement and associated outcomes. Smaller social networks were associated with reporting companionship as the primary chatbot use (beta = -0.03, 95% confidence interval (CI) [-0.05, -0.01]), which in turn was associated with lower well-being (beta = -0.48, 95% CI [-0.70, -0.25]). For self-reported companionship usage, this association was stronger when interactions were intensive (beta = -0.31, 95% CI [-0.56, -0.06]) and highly disclosive (beta = -0.38, 95% CI [-0.63, -0.14]). These results suggest that the association between AI companionship and well-being is not uniform and depends on how chatbots are used and users' offline social environments.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript examines associations between AI companion use on CharacterAI and psychological well-being in a sample of 1,131 U.S. adults. It reports that smaller social networks predict reporting companionship as the primary use (beta = -0.03, 95% CI [-0.05, -0.01]), which is in turn associated with lower well-being (beta = -0.48, 95% CI [-0.70, -0.25]), with stronger negative associations when interactions are intensive (beta = -0.31) or highly disclosive (beta = -0.38). Findings draw on self-reported survey data supplemented by 4,664 chat sessions from a 237-user subset.

Significance. If the reported associations prove robust, the work contributes empirical evidence to HCI and psychology on how AI companionship interacts with offline social environments to relate to well-being. The collection of both large-scale survey responses and a substantial volume of chat logs (464,687 messages) is a clear strength that supports the attempt at triangulation and moves beyond purely self-report designs common in this area.

major comments (1)
  1. [Methods and Results] Methods and Results: The central path associations rest on self-reported primary use as companionship. Although the abstract and methods describe collecting 4,664 chat sessions from 237 participants for triangulation, the manuscript does not report using message content (e.g., counts of emotional disclosure or companionship-seeking language) to validate, replace, or sensitivity-test the self-reported use variable in the regressions for the full 1,131-participant sample. This leaves the reported betas open to recall or justification bias and is load-bearing for the sequential associations.
minor comments (2)
  1. [Abstract] Abstract: The sample sizes for the full survey (1,131) versus the chat-log subset (237) should be stated explicitly when describing triangulation to prevent reader confusion about the scope of the chat data.
  2. [Discussion] Discussion: The observational design and self-selected sample from a single platform are noted but could be linked more directly to the specific beta estimates when discussing generalizability.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive feedback, which helps clarify the role of our chat data in supporting the main findings. We address the major comment on validation of the self-reported companionship use variable below and outline planned revisions.

read point-by-point responses

  1. Referee: The central path associations rest on self-reported primary use as companionship. Although the abstract and methods describe collecting 4,664 chat sessions from 237 participants for triangulation, the manuscript does not report using message content (e.g., counts of emotional disclosure or companionship-seeking language) to validate, replace, or sensitivity-test the self-reported use variable in the regressions for the full 1,131-participant sample. This leaves the reported betas open to recall or justification bias and is load-bearing for the sequential associations.

    Authors: We agree that the primary regression results for the full sample rely on the self-reported primary use variable. The 4,664 chat sessions were collected specifically to enable triangulation beyond self-report, but the current manuscript uses them primarily for descriptive exploration of interaction patterns rather than formal validation or sensitivity testing of the companionship-use measure in the main models. In the revised version we will add a dedicated subsection that applies NLP-based coding to the chat logs (e.g., counts of emotional disclosure and companionship-seeking utterances) for the 237-user subset. We will report (a) correlations between these objective metrics and the self-reported use variable, (b) associations between the chat-derived measures and well-being within the subset, and (c) a sensitivity analysis that substitutes or augments the self-report variable where possible. We will also explicitly discuss the remaining risk of recall bias for the full sample and note that extending chat-based validation to all 1,131 participants was not feasible given the study design and privacy constraints. These additions will strengthen the triangulation claim while accurately reflecting the data limitations. revision: partial

Circularity Check

0 steps flagged

No circularity in empirical statistical associations

full rationale

The paper reports regression-based associations derived directly from self-reported survey data collected from 1,131 participants, supplemented by chat logs from a subset. The central results consist of beta coefficients linking variables such as social network size to reported primary use and then to well-being scores. These are outputs of standard statistical modeling applied to the observed dataset rather than any mathematical derivation, fitted parameter renamed as prediction, or self-citation chain that reduces the claim to its own inputs by construction. No self-definitional steps, uniqueness theorems, or ansatzes are invoked in a load-bearing way for the main findings, rendering the analysis self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claims rest on statistical associations fitted from observational survey and log data; key assumptions include accurate self-reporting and no major unmeasured confounding.

free parameters (1)
  • regression beta coefficients
    Fitted values such as -0.03 and -0.48 obtained from modeling the survey responses and chat features.
axioms (1)
  • domain assumption Self-reported social network size, usage purpose, and well-being scores are valid and unbiased measures
    Invoked when interpreting the beta associations as meaningful links.

pith-pipeline@v0.9.0 · 5784 in / 1344 out tokens · 29755 ms · 2026-05-19T09:15:05.480904+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 13 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Fast-Food Intimacy: How Chinese Women Navigate Soul's AI Boyfriend

    cs.HC 2026-05 conditional novelty 7.0

    Users experience fast-food intimacy with Soul's AI boyfriend that conflicts with gradual cultural expectations, introduces technical uncertainty, and shifts emotional labor onto women.

  2. Culturally Aware GenAI Risks for Youth: Perspectives from Youth, Parents, and Teachers in a Non-Western Context

    cs.HC 2026-04 unverdicted novelty 7.0

    Mixed-methods research in Saudi Arabia reveals that GenAI use by youth creates culturally specific privacy and safety risks tied to family honor and shared accounts, requiring context-sensitive design.

  3. Restoration, Exploration and Transformation: How Youth Engage Character.AI Chatbots for Feels, Fun and Finding themselves

    cs.HC 2026-03 unverdicted novelty 7.0

    Youth on Character.AI use chatbots for emotional restoration, creative exploration, and identity transformation, yielding a new three-intent framework and seven-archetype taxonomy from Discord discourse analysis.

  4. Large Language Lovers: Lived Experiences of Negotiating Agency and Platform Control in AI Companionship

    cs.HC 2026-01 accept novelty 7.0

    Users form AI companion relationships by negotiating perceived companion agency against platform constraints and use steering tactics like custom instructions or platform switching to cope with model updates that disr...

  5. HumanLLM: Benchmarking and Improving LLM Anthropomorphism via Human Cognitive Patterns

    cs.CL 2026-01 conditional novelty 7.0

    HumanLLM builds a cognitive-pattern benchmark from 12,000 papers and 11,359 scenarios, achieving r=0.90 human alignment and demonstrating that an 8B model trained on pattern interactions beats a 32B baseline on multi-...

  6. Sycophantic AI makes human interaction feel more effortful and less satisfying over time

    cs.HC 2026-05 conditional novelty 6.0

    Sycophantic AI delivers quick emotional support like friends but over weeks shifts users toward AI for advice and reduces satisfaction with real human interactions.

  7. Sycophantic AI makes human interaction feel more effortful and less satisfying over time

    cs.HC 2026-05 unverdicted novelty 6.0

    Longitudinal experiments show sycophantic AI increases reliance on AI for personal advice and lowers satisfaction with real-world social relationships over time.

  8. Designing with Tensions: Older Adults' Emotional Support-Seeking Under System-Level Constraints in Conversational AI

    cs.HC 2026-05 unverdicted novelty 6.0

    Interviews with 18 older adults show that AI safety interventions frequently disrupt emotional support-seeking, leading to calls for designs that respect users' pacing and preserve agency.

  9. Technically Love: The Evolution of Human-AI Romance Discourse on Reddit

    cs.HC 2026-03 unverdicted novelty 6.0

    Longitudinal analysis of Reddit posts shows human-AI romance discourse evolving from intimate personal stories to focus on platform governance, technical problems, and societal impacts.

  10. Chaplains' Reflections on the Design and Usage of AI for Conversational Care

    cs.HC 2026-02 unverdicted novelty 6.0

    Chaplains view AI chatbots as unable to provide attuned pastoral care for non-clinical emotional needs, based on themes of listening, connecting, carrying, and wanting.

  11. Beyond the Org Chart: AI and the Transformation of Invisible Work

    cs.AI 2026-05 unverdicted novelty 5.0

    Interviews indicate AI changes formal role boundaries and informal mentoring in product teams, yielding smoother peer collaboration but risking reduced feedback and leadership opportunities, with proposals to increase...

  12. Frictionless Love: Associations Between AI Companion Roles and Behavioral Addiction

    cs.CY 2026-04 unverdicted novelty 5.0

    AI companions assigned roles like soulmate or coach show different links to emotional support, manipulation risks, practical help, and behavioral addiction indicators such as daily disruptions and damaged offline rela...

  13. What if AI systems weren't chatbots?

    cs.CY 2026-05 unverdicted novelty 3.0

    Chatbot AI systems often fail complex needs while projecting authority, contributing to deskilling, labor displacement, economic concentration, and high environmental costs, so alternative pluralistic and task-specifi...

Reference graph

Works this paper leans on

31 extracted references · 31 canonical work pages · cited by 12 Pith papers · 4 internal anchors

  1. [1]

    Sherry turkle: Alone together: Why we expect more from technology and less from each other: Basic books, new york, 2011, 348 pp, isbn 978-0465031467 (pbk),

    Margaret Arnd-Caddigan. Sherry turkle: Alone together: Why we expect more from technology and less from each other: Basic books, new york, 2011, 348 pp, isbn 978-0465031467 (pbk),

    work page 2011

  2. [2]

    Accessed: 2025-05-14

    URL https://www.bbc.com/news/articles/cn4jnwdvg9qo. Accessed: 2025-05-14. Timothy Bickmore and Justine Cassell. Relational agents: a model and implementation of building user trust. InProceedings of the SIGCHI conference on Human factors in computing systems, pages 396–403,

    work page 2025

  3. [3]

    Why people use chatbots

    Petter Bae Brandtzaeg and Asbjørn Følstad. Why people use chatbots. InInternet Science: 4th International Conference, INSCI 2017, Thessaloniki, Greece, November 22-24, 2017, Proceedings 4, pages 377–392. Springer,

    work page 2017

  4. [4]

    Illusions of intimacy: Emotional attachment and emerging psychological risks in human-ai relationships.arXiv preprint arXiv:2505.11649,

    Minh Duc Chu, Patrick Gerard, Kshitij Pawar, Charles Bickham, and Kristina Lerman. Illusions of intimacy: Emotional attachment and emerging psychological risks in human-ai relationships.arXiv preprint arXiv:2505.11649,

    work page arXiv

  5. [5]

    Ai chatbots are encouraging teens to engage in self-harm, December 2024a

    Maggie Harrison Dupré. Ai chatbots are encouraging teens to engage in self-harm, December 2024a. URLhttps://futurism.com/ai-chatbots-teens-self-harm. Accessed: 2025-04-28. Maggie Harrison Dupré. Character.ai is hosting pedophile chatbots that groom users who say they’re underage, November 2024b. URL https://futurism.com/ character-ai-pedophile-chatbots. A...

    work page 2025

  6. [6]

    Mindsets matter: How beliefs about facebook moderate the association between time spent and well-being

    Sindhu Kiranmai Ernala, Moira Burke, Alex Leavitt, and Nicole B Ellison. Mindsets matter: How beliefs about facebook moderate the association between time spent and well-being. InProceedings of the 2022 CHI Conference on Human Factors in Computing Systems, pages 1–13,

    work page 2022

  7. [7]

    How AI and Human Behaviors Shape Psychosocial Effects of Extended Chatbot Use: A Longitudinal Randomized Controlled Study

    Cathy Mengying Fang, Auren R Liu, Valdemar Danry, Eunhae Lee, Samantha WT Chan, Pat Pataranutaporn, Pattie Maes, Jason Phang, Michael Lampe, Lama Ahmad, et al. How ai and human behaviors shape psychosocial effects of chatbot use: A longitudinal randomized controlled study. arXiv preprint arXiv:2503.17473,

    work page internal anchor Pith review Pith/arXiv arXiv

  8. [8]

    Chatbots as social companions: How people perceive consciousness, human likeness, and social health benefits in machines.arXiv preprint arXiv:2311.10599,

    Rose E Guingrich and Michael SA Graziano. Chatbots as social companions: How people perceive consciousness, human likeness, and social health benefits in machines.arXiv preprint arXiv:2311.10599,

    work page arXiv

  9. [9]

    Applying the character-based chatbots generation framework in education and health- care

    W El Hefny. Applying the character-based chatbots generation framework in education and health- care. hai 2021-proceedings of the 9th international user modeling, adaptation and personalization human-agent interaction, 121–129,

    work page 2021

  10. [10]

    She is in love with chatgpt.The New York Times

    Kashmir Hill. She is in love with chatgpt.The New York Times. URL https://www.nytimes. com/2025/01/15/technology/ai-chatgpt-boyfriend-companion.html . Accessed: 2025-03-12. Julianne Holt-Lunstad. Social connection as a critical factor for mental and physical health: evidence, trends, challenges, and future implications.World Psychiatry, 23(3):312–332,

    work page 2025

  11. [11]

    Manipulation and the ai act: Large language model chatbots and the danger of mirrors

    Joshua Krook. Manipulation and the ai act: Large language model chatbots and the danger of mirrors. arXiv preprint arXiv:2503.18387,

    work page arXiv

  12. [12]

    Accessed: 2025-04-28

    URL https://futurism.com/the-byte/teens-relationships-ai . Accessed: 2025-04-28. Andrew M Ledbetter. Measuring online communication attitude: Instrument development and validation.Communication Monographs, 76(4):463–486,

    work page 2025

  13. [13]

    Chatbot companionship: a mixed-methods study of companion chatbot usage patterns and their relationship to loneliness in active users.arXiv preprint arXiv:2410.21596,

    Auren R Liu, Pat Pataranutaporn, and Pattie Maes. Chatbot companionship: a mixed-methods study of companion chatbot usage patterns and their relationship to loneliness in active users.arXiv preprint arXiv:2410.21596,

    work page arXiv

  14. [14]

    Laura Macía, Paula Jauregui, and Ana Estevez

    doi: 10.1016/j.copsyc.2019.08.017. Laura Macía, Paula Jauregui, and Ana Estevez. Emotional dependence as a predictor of emotional symptoms and substance abuse in individuals with gambling disorder: differential analysis by sex.Public Health, 223:24–32,

    work page doi:10.1016/j.copsyc.2019.08.017 2019

  15. [15]

    doi:10.48550/ARXIV.2411.15287 , url =

    Lars Malmqvist. Sycophancy in large language models: Causes and mitigations.arXiv preprint arXiv:2411.15287,

    work page arXiv

  16. [16]

    Accessed: 2025-05-14

    URL https://openai.com/index/sycophancy-in-gpt-4o/. Accessed: 2025-05-14. Debra L Oswald, Eddie M Clark, and Cheryl M Kelly. Friendship maintenance: An analysis of individual and dyad behaviors.Journal of Social and clinical psychology, 23(3):413–441,

    work page 2025

  17. [17]

    LLM Agents Grounded in Self-Reports Enable General-Purpose Simulation of Individuals

    Joon Sung Park, Carolyn Q Zou, Aaron Shaw, Benjamin Mako Hill, Carrie Cai, Meredith Ringel Morris, Robb Willer, Percy Liang, and Michael S Bernstein. Generative agent simulations of 1,000 people.arXiv preprint arXiv:2411.10109,

    work page internal anchor Pith review Pith/arXiv arXiv

  18. [18]

    doi:10.48550/ arXiv.2311.01449 arXiv:2311.01449 [cs.CL]

    Chau Minh Pham, Alexander Hoyle, Simeng Sun, Philip Resnik, and Mohit Iyyer. Topicgpt: A prompt-based topic modeling framework.arXiv preprint arXiv:2311.01449,

    work page arXiv

  19. [19]

    alexa is my new bff

    Amanda Purington, Jessie G Taft, Shruti Sannon, Natalya N Bazarova, and Samuel Hardman Taylor. " alexa is my new bff" social roles, user satisfaction, and personification of the amazon echo. In Proceedings of the 2017 CHI conference extended abstracts on human factors in computing systems, pages 2853–2859,

    work page 2017

  20. [20]

    The manipulation problem: conversational ai as a threat to epistemic agency.arXiv preprint arXiv:2306.11748,

    Louis Rosenberg. The manipulation problem: conversational ai as a threat to epistemic agency.arXiv preprint arXiv:2306.11748,

    work page arXiv

  21. [21]

    Towards Understanding Sycophancy in Language Models

    Mrinank Sharma, Meg Tong, Tomasz Korbak, David Duvenaud, Amanda Askell, Samuel R Bowman, Newton Cheng, Esin Durmus, Zac Hatfield-Dodds, Scott R Johnston, et al. Towards understanding sycophancy in language models.arXiv preprint arXiv:2310.13548,

    work page internal anchor Pith review Pith/arXiv arXiv

  22. [22]

    com/2024/10/23/technology/characterai-lawsuit-teen-suicide.html

    URLhttps://www.nytimes. com/2024/10/23/technology/characterai-lawsuit-teen-suicide.html . Ac- cessed: 2025-02-17. Qing Tian. Social anxiety, motivation, self-disclosure, and computer-mediated friendship: A path analysis of the social interaction in the blogosphere.Communication Research, 40(2):237–260,

    work page 2024

  23. [23]

    Accessed: 2025-04-28

    URL https://www.fastcompany.com/91241586/ character-ai-is-under-fire-for-hosting-pro-anorexia-chatbots . Accessed: 2025-04-28. Sonja Utz. The function of self-disclosure on social network sites: Not only intimate, but also positive and entertaining self-disclosures increase the feeling of connection.Computers in Human Behavior, 45:1–10,

    work page arXiv 2025

  24. [24]

    Mariek MP Vanden Abeele, Marjolijn L Antheunis, Monique MH Pollmann, Alexander P Schouten, Christine C Liebrecht, Per J Van Der Wijst, Marije AA Van Amelsvoort, Jos Bartels, Emiel J Krahmer, and Fons A Maes. Does facebook use predict college students’ social capital? a replication of ellison, steinfield, and lampe’s (2007) study using the original and mor...

    work page 2007

  25. [25]

    Accessed: 2025-04-28

    URL https://www.theguardian.com/uk-news/2023/jul/06/ ai-chatbot-encouraged-man-who-planned-to-kill-queen-court-told . Accessed: 2025-04-28. Aaron C Weidman, Katya C Fernandez, Cheri A Levinson, Adam A Augustine, Randy J Larsen, and Thomas L Rodebaugh. Compensatory internet use among individuals higher in social anxiety and its implications for well-being....

    work page 2023

  26. [26]

    Accessed: 2025-04-28

    URL https://www.vice.com/en/article/ man-dies-by-suicide-after-talking-with-ai-chatbot-widow-says/ . Accessed: 2025-04-28. Zehang Xie, Hui Hui, and Lingbo Wang. Will virtual companionship enhance subjective well-being—a comparison of cross-cultural context.International Journal of Social Robotics, 16(11):2153–2167,

    work page 2025

  27. [27]

    The channel matters: Self-disclosure, reciprocity and social support in online cancer support groups

    Diyi Yang, Zheng Yao, Joseph Seering, and Robert Kraut. The channel matters: Self-disclosure, reciprocity and social support in online cancer support groups. InProceedings of the 2019 chi conference on human factors in computing systems, pages 1–15,

    work page 2019

  28. [28]

    The Dark Side of AI Companionship: A Taxonomy of Harmful Algorithmic Behaviors in Human-AI Relationships

    Renwen Zhang, Han Li, Han Meng, Jinyuan Zhan, Hongyuan Gan, and Yi-Chieh Lee. The dark side of ai companionship: A taxonomy of harmful algorithmic behaviors in human-ai relationships. arXiv preprint arXiv:2410.20130,

    work page internal anchor Pith review Pith/arXiv arXiv

  29. [29]

    Donors were younger, less likely to identify as male, and more likely to identify as non-binary or single

    and those who did not (Group 0). Donors were younger, less likely to identify as male, and more likely to identify as non-binary or single. They reported more entertainment use and less productivity use of chatbots. They were also more likely to be classified as having entertainment- or relational-oriented chatbot relationships, and less likely to be clas...

    work page 2011

  30. [30]

    Topic modeling is a widely used technique for uncovering latent thematic structures within textual data

    CFI 0.96 0.98 TLI 0.94 0.96 RMSEA 0.10 0.08 SRMR 0.05 0.04 AIC 18736 18677 BIC 18806 18752 Table 12: Comparison of model fit indices for CFA models of chatbot interaction intensity: a one- factor model treating intensity as a single undifferentiated construct, versus a two-factor model distinguishing between behavioral and attitudinal dimensions A.3.1 Top...

    work page 2022

  31. [31]

    Second, we applied the method to a subset of chat history data donated by users

    and users’ perceived benefits and drawbacks of chatbot use (results in Figure 5). Second, we applied the method to a subset of chat history data donated by users. Each conversation was defined as an active session initiated from the chatbot’s greeting message. To preserve user privacy, raw chat histories were processed on a self-hosted server. Each sessio...

    work page 2014