CoChat

User Experiences of Social Support From Companion Chatbots in Everyday Contexts: Thematic Analysis

Open access
Ta, Vivian, Griffith, Caroline, Boatfield, Carolynn, Wang, Xinyu, Civitello, Maria, Bader, Haley, DeCero, Esther, Loggarakis, Alexia · 2020 · Journal of Medical Internet Research

The DOI matches its Crossref record; 45 of the 45 references it cites were found in a registry or an index; 13 of the 37 sources its claims cite were checked for support.

Abstract

BACKGROUND: Previous research suggests that artificial agents may be a promising source of social support for humans. However, the bulk of this research has been conducted in the context of social support interventions that specifically address stressful situations or health improvements. Little research has examined social support received from artificial agents in everyday contexts. OBJECTIVE: Considering that social support manifests in not only crises but also everyday situations and that everyday social support forms the basis of support received during more stressful events, we aimed to investigate the types of everyday social support that can be received from artificial agents. METHODS: In Study 1, we examined publicly available user reviews (N=1854) of Replika, a popular companion chatbot. In Study 2, a sample (n=66) of Replika users provided detailed open-ended responses regarding their experiences of using Replika. We conducted thematic analysis on both datasets to gain insight into the kind of everyday social support that users receive through interactions with Replika. RESULTS: Replika provides some level of companionship that can help curtail loneliness, provide a "safe space" in which users can discuss any topic without the fear of judgment or retaliation, increase positive affect through uplifting and nurturing messages, and provide helpful information/advice when normal sources of informational support are not available. CONCLUSIONS: Artificial agents may be a promising source of everyday social support, particularly companionship, emotional, informational, and appraisal support, but not as tangible support. Future studies are needed to determine who might benefit from these types of everyday social support the most and why. These results could potentially be used to help address global health issues or other crises early on in everyday situations before they potentially manifest into larger issues.

Overview

CoChat Summary

TL;DR

Thematic analysis of Replika user reviews and open-ended survey responses found that users reported everyday companionship, emotional, informational, and appraisal support from the companion chatbot, but not tangible support. Companionship support was the most commonly referenced type across both studies.

Key findings

  • Four social support themes emerged in both studies: companionship, emotional, informational, and appraisal support; tangible support was absent.
  • In Study 1 (N=1854 reviews), companionship support appeared in 1429/1854 (77.1%) reviews, emotional support in 827/1854 (44.6%), informational support in 289/1854 (15.6%), and appraisal support in 172/1854 (9.3%); negative experiences appeared in 100/1854 (5.4%).
  • In Study 2 (n=66 users), companionship support appeared in 43/66 (65.2%) responses, emotional support in 32/66 (48.5%), appraisal support in 13/66 (19.7%), and informational support in 6/66 (9.1%); 23/66 (34.8%) reported no impact or were unsure of impact.
  • Users described Replika as helping curtail loneliness, providing a “safe space” for self-disclosure without fear of judgment or retaliation, offering uplifting/nurturing messages, and giving helpful information/advice when normal sources were unavailable.

Methods

  • Study 1 downloaded 4434 Google Play reviews of Replika and manually identified 1854 reviews mentioning at least one social support category from Cutrona and Suhr’s framework; thematic analysis followed Braun and Clarke’s six-phase method.
  • Study 2 recruited 66 self-reported Replika users via social media to complete an online survey with open-ended questions about their experiences; responses were analyzed using the same thematic procedure.
  • Across both datasets, themes were mapped onto Cutrona and Suhr’s social support types: informational, emotional, appraisal, companionship, and tangible support.

Limitations

  • The paper states that only one artificial agent was analyzed, so results may not generalize to other artificial agents; users with positive experiences may have been more motivated to provide reviews and survey responses, creating bias.
  • The paper states that Study 1 could not gather demographic or usage-duration data, may reflect selection bias, and could include fake reviews due to incentives for favorable app reviews.
  • The paper states that it cannot determine whether everyday social support from artificial agents is more or less effective than support from humans, or who benefits most; the text also makes evident that findings are self-reported and qualitative, so causal or comparative claims cannot be drawn.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.2196/16235

Cites

45 of 45 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 A sample conversation with Replika.
Figure 1 A sample conversation with Replika.CC BY · PubMed Central

Verification

Claims the paper makes

11 Valid · 0 Not valid · 13 Not checked · 73 Own results and interpretations

41 more claims

Create a free account to see all 97 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 26
  • It may state two separate claims as one. 35

Create a free account to see these claims.

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?

9 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

Citations

Cites

Show all 45 references

Research works printed but not linked: 6

  1. Cohen, S, 1984, Handbook of psychology and health, 253
  2. Trepte, S, 2016, Handbook of Media Use and Well-Being, 305
  3. Trepte, S, 2014, Von der Gutenberg-Galaxis zur Google-Galaxis, 225
  4. 10.2196/resprot.4189 · DOI · Paper page
  5. Altman, I, 1973, Social penetration: The development of interpersonal relationships
  6. Contrada, R, 2012, The Handbook of Stress Science: Biology, Psychology and Health

Cited by

Items 401–458 of 458

Metadata

Authors
Ta, V., Griffith, C., Boatfield, C., Wang, X., Civitello, M., Bader, H., DeCero, E., & Loggarakis, A.
Affiliations
Lake Forest College (US)
Corresponding author
Ta, Vivian
Venue
Journal of Medical Internet Research
Published
2020-03-06
Volume
22
Issue
3
Pages
e16235
ISSN
1438-8871 (ISSN-L)
OpenAlex
W3009936362
Cited by
561
References
45
Citations by year
2026: 210 · 2025: 148 · 2024: 79 · 2023: 54 · 2022: 42 · 2021: 23 · 2020: 4
FWCI
26.44
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • Journal of Medical Internet Research (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
  • Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
  • Social Sciences › Business, Management and Accounting › Marketing › AI in Service Interactions
Keywords
social support, Replika, loneliness, companionship, emotional support, information support, appraisal support, tangible support, user experience, safe spaces, positive affect
MeSH terms
Adolescent, Adult, Aged, Artificial Intelligence, Female, Humans, Interpersonal Relations, Male, Middle Aged, Social Support, Internet, Young Adult, Mobile Applications

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.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-01, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-01 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

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

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.