CoChat
Sign Up
Cited by 241Digital Mental Health InterventionsSocial Robot Interaction and HRIArtificial Intelligence in Healthcare and Education

“It happened to be the perfect thing”: experiences of generative AI chatbots for mental health

Open access
Siddals, Steven, Torous, John, Coxon, Astrid·2024·npj Mental Health Research
DOI 10.1038/s44184-024-00097-4PMC PMC11514308

Published version of a preprint

Open this paper in CoChat — free account

Abstract

The global mental health crisis underscores the need for accessible, effective interventions. Chatbots based on generative artificial intelligence (AI), like ChatGPT, are emerging as novel solutions, but research on real-life usage is limited. We interviewed nineteen individuals about their experiences using generative AI chatbots for mental health. Participants reported high engagement and positive impacts, including better relationships and healing from trauma and loss. We developed four themes: (1) a sense of ‘ emotional sanctuary’ , (2) ‘ insightful guidance’ , particularly about relationships, (3) the ‘ joy of connection ’, and (4) comparisons between the ‘ AI therapist ’ and human therapy. Some themes echoed prior research on rule-based chatbots, while others seemed novel to generative AI. Participants emphasised the need for better safety guardrails, human-like memory and the ability to lead the therapeutic process. Generative AI chatbots may offer mental health support that feels meaningful to users, but further research is needed on safety and effectiveness.

AI snapshot

Written from the paper’s full text.

TL;DR

In semi-structured interviews with 19 people using generative AI chatbots for mental health, participants reported high engagement and positive impacts including improved relationships, healing from trauma and loss, and improved mood, organized into four themes: emotional sanctuary, insightful guidance, joy of connection, and comparisons with the “AI therapist.” They also described frustrations with safety guardrails, listening/response quality, and lack of memory or therapeutic leadership, leading the authors to call for more research on safety and effectiveness.

Key findings

  • Nineteen participants (12 male, 7 female; ages 17–60; eight countries) reported mostly positive impacts, including improved relationships, healing from trauma and loss, improved mood, and help with existing therapeutic journeys; some described the impact as life-changing, while one participant reported negligible impact.
  • Four overarching themes were developed: (1) “Emotional sanctuary,” (2) “Insightful guidance,” (3) “Joy of connection,” and (4) “The AI therapist?”; most participants used Pi, several used ChatGPT, and a few used Copilot, Kindroid, ChatMind and others, with a majority using generative AI chatbots at least several times a week.
  • Participants valued understanding, validation, non-judgmental availability, relationship advice, reframing, and companionship, but many also reported frustration with irrelevant or overly long responses, advice offered before they felt heard, and safety guardrails that felt rejecting or arbitrary.
  • Many participants missed human-like memory and the chatbot’s ability to lead the therapeutic process; some used chatbots creatively for role-play, imagery, and fictional characters, and several compared chatbots positively and negatively with human therapy.

Methods

  • Qualitative study using semi-structured interviews and reflexive thematic analysis; 19 participants were recruited by convenience sampling via user forums (Pi, Reddit, IFS guide app), King’s College London students/staff, and LinkedIn.
  • Inclusion required at least three separate conversations with an LLM-based generative AI chatbot on mental health/wellbeing topics, each lasting at least 20 minutes, age over 16, and comfort being interviewed in English; 35 individuals signed consent and 19 attended interviews.
  • Interviews were conducted online between 10 January and 16 March 2024, lasted 49–112 minutes, and were analyzed inductively following Braun & Clarke’s reflexive thematic analysis, producing around 600 codes and four themes.

Limitations

  • Stated: convenience sampling meant many populations were not represented; most participants lived in high-income countries, were tech-savvy and well-educated, focused on milder mental health conditions, and all self-selected, potentially biasing toward positive experiences and missing those for whom the technology did not work.
  • Stated: reflexive thematic analysis involves subjectivity, especially as conducted by a sole researcher (SS), though AC reviewed codes and themes to support rigour and validity.
  • Evident from the text: the qualitative design captures perceptions rather than clinical outcomes, so it cannot establish effectiveness or safety, and no quantitative symptom, impairment, or relapse measures were reported.

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

Integrity signals

Registry

Crossref · 10.1038/s44184-024-00097-4

Cites

Statistics

Nothing to recompute.

Claims the paper makes

0 of 67 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Several participants mentioned using generative AI for role-play, whether to explore different, healthier ways of relating, to prepare for conflict, or to experience a healing conversation.

    “Several participants mentioned using generative AI for role-play, whether to explore different, healthier ways of relating, to prepare for conflict, or in Isabel’s case, to experience a healing conversation that her father would be unlikely to offer –”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.
  • Findingown data

    Several participants found the chatbot's value limited by its inability to take the lead in the therapeutic process.

    “Several participants found the chatbot’s value limited by its inability to take the lead in the therapeutic process, either to help the client through intense emotions –”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.
  • Findingown data

    While others experienced being proactively challenged, in a supportive way.

    “While others experienced being proactively challenged, in a supportive way –”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.
  • Findingown data

    For many participants, generative AI chatbots do not match human empathy and connection.

    “For many participants however, generative AI chatbots do not match human empathy and connection –”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.
  • Findingown data

    For some participants, generative AI chatbots were a valuable tool to augment therapy.

    “Finally, for clinicians, our study found that for some participants, generative AI chatbots were a valuable tool to augment therapy.”
    Quote in the paperSubstantial claim
  • May combine two claims. 50
  • Classified as probably not a substantial claim. 31

  • May combine two claims.

    Findingown data

    Participants reported high engagement and positive impacts, including better relationships and healing from trauma and loss.

    “Participants reported high engagement and positive impacts, including better relationships and healing from trauma and loss.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    Participants emphasised the need for better safety guardrails, human-like memory and the ability to lead the therapeutic process.

    “Participants emphasised the need for better safety guardrails, human-like memory and the ability to lead the therapeutic process.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    Four themes were developed: (1) a sense of 'emotional sanctuary', (2) 'insightful guidance', particularly about relationships, (3) the 'joy of connection', and (4) comparisons between the 'AI therapist' and human therapy.

    “We developed four themes: (1) a sense of ‘emotional sanctuary’, (2) ‘insightful guidance’, particularly about relationships, (3) the ‘joy of connection’, and (4) comparisons between the ‘AI therapist’ and human therapy.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The sample of nineteen participants is sufficient to develop meaningful themes about experiences with generative AI chatbots.
    • Rests on an assumption: The themes developed from the interviews accurately represent the participants' experiences.

  • May combine two claims.

    Findingown data

    For some, the experience of safety guardrails felt unpleasant, limiting and awkward, while for others, encountering guardrails felt like a rejection in a time of need.

    “For some, the experience felt unpleasant, limiting and awkward, while for others, encountering guardrails felt like a rejection in a time of need –”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    Most participants used Pi (from Inflection), several used ChatGPT (OpenAI), and a few used Copilot (Microsoft), Kindroid (Kindroid), ChatMind (VOS) and others.

    “Most participants used Pi (from Inflection), several used ChatGPT (OpenAI), and a few used Copilot (Microsoft), Kindroid (Kindroid), ChatMind (VOS) and others.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    While other participants reported a high level of overall trust in the chatbot's judgement.

    “While other participants reported a high level of overall trust in its judgement –”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    In addition to creating space for emotions, most participants also valued the guidance and advice they received, especially on relationships.

    “In addition to creating space for emotions, most participants also valued the guidance and advice they received, especially on relationships.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    Most participants reported that their use of generative AI chatbots had impacted their lives positively, in various ways, including improved relationships, healing from trauma and loss, improved mood, as well as by helping their existing therapeutic journeys.

    “Most participants reported that their use of generative AI chatbots had impacted their lives positively, in various ways, including improved relationships, healing from trauma and loss, improved mood, as well as by helping their existing therapeutic journeys.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    Several participants had also experienced rule-based mental health apps and commented on how they offer a less satisfying user experience.

    “Several participants had also experienced rule-based mental health apps and commented on how they offer a less satisfying user experience –”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    For one participant the impact of generative AI chatbots was negligible.

    “While for one participant the impact was negligible –”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    A variety of topics brought participants to use generative AI chatbots, including anxiety, depression, stress, conflict, dealing with loss and romantic relationships.

    “A variety of topics brought participants to use generative AI chatbots, including anxiety, depression, stress, conflict, dealing with loss and romantic relationships.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    Some participants mentioned that the chatbot helped them see the other person's perspective in conflict, or coached them through difficult relationship situations.

    “Some participants mentioned that it helped them see the other person’s perspective in conflict, or coached them through difficult relationship situations –”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    A range of positive impacts were reported, including improved mood, reduced anxiety, healing from trauma and loss, and improved relationships, that, for some, were considered life changing.

    “A range of positive impacts were reported, including improved mood, reduced anxiety, healing from trauma and loss, and improved relationships, that, for some, were considered life changing.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    Most participants in the study used the Pi chatbot.

    “the Pi chatbot used by most of the participants in our study”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Generative AI chatbot usage among the study's participants was diverse, complex and personalised.

    “These simple questions may not yield clear answers, as our study shows that generative AI chatbot usage is diverse, complex and personalised, and moreover, constantly evolving as the underlying technology improves.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Many participants reported high frequency of use and most reported high levels of satisfaction.

    “Many participants reported high frequency of use and most reported high levels of satisfaction.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    The level of trust in the chatbot's guidance was mixed, with many participants reporting scepticism, or experiencing hallucination or unsatisfying advice.

    “The level of trust in the chatbot’s guidance was mixed, with many participants reporting scepticism, or experiencing hallucination or unsatisfying advice –”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    Several found it helpful to augment their therapy with chatbot usage, with mixed reactions from the therapist in some cases.

    “Several found it helpful to augment their therapy with chatbot usage, with mixed reactions from the therapist in some cases –”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • Classified as probably not a substantial claim. May combine two claims.

    Findingown data

    Four overarching themes were developed: (1) 'Emotional sanctuary', (2) 'Insightful guidance', (3) 'Joy of connection', and (4) 'The AI therapist?'.

    “Four overarching themes were developed, summarised in Fig. 3 and shown with subthemes in Supplementary Figure 1: (1) ‘Emotional sanctuary’, (2) ‘Insightful guidance’, (3) ‘Joy of connection’, and (4) ‘The AI therapist?’”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The sample of nineteen participants is sufficient to develop meaningful themes about experiences with generative AI chatbots.
    • Rests on an assumption: The themes developed from the interviews accurately represent the participants' experiences.

  • May combine two claims.

    Findingown data

    Participants told us that generative AI feels like an emotional sanctuary, offers insightful guidance, can be a joy to connect with and can connect people to others, and bears comparison with human therapy (positively and negatively).

    “Participants told us that generative AI feels like an emotional sanctuary, offers insightful guidance, can be a joy to connect with and can connect people to others, and bears comparison with human therapy (positively and negatively).”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.
    • Rests on an assumption: The sample of nineteen participants is sufficient to develop meaningful themes about experiences with generative AI chatbots.
    • Rests on an assumption: The themes developed from the interviews accurately represent the participants' experiences.

  • May combine two claims.

    Findingown data

    Despite overall positive experiences, a majority of participants also experienced frustration with how well the chatbots listen and respond, for example, with irrelevant or overly long responses, or offering advice before the user felt fully heard.

    “Despite overall positive experiences, a majority of participants also experienced frustration with how well the chatbots listen and respond, for example, with irrelevant or overly long responses, or offering advice before the user felt fully heard –”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    Several participants experienced meaningful crisis support from generative AI, as long as guardrails were not triggered.

    “Several participants experienced meaningful crisis support from generative AI, as long as guardrails were not triggered.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    A majority of participants experienced generative AI chatbots as understanding, validating, patient, kind, non-judgmental, always available and expecting nothing in return.

    “A majority of participants experienced generative AI chatbots as understanding, validating patient, kind, non-judgmental, always available and expecting nothing in return.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    Many participants turned to chatbots because therapy was not an option, either due to cost and availability, or because therapy did not give them the help they were looking for.

    “Many participants turned to chatbots because therapy was not an option, either due to cost and availability, or because therapy did not give them the help they were looking for.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    A majority of participants used generative AI chatbots at least several times a week.

    “A majority of participants used generative AI chatbots at least several times a week.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Findingown data

    Participants reported high engagement, positive impacts, and novel experiences in comparison with existing digital mental health interventions.

    “with participants reporting high engagement, positive impacts, and novel experiences in comparison with existing DMHIs”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Meta-analyses indicate that the therapeutic effects of rule-based AI chatbots are small and not sustained over time, and users report frustration with responses that feel empty, generic, nonsensical, repetitive and constrained.

    “Meta-analyses indicate that the therapeutic effects are small and not sustained over time24, and users report frustration with responses that feel empty, generic20, nonsensical, repetitive and constrained19–22.”
    Quote in the paperSubstantial claim0 of 5 sources checked
    • cites [24] He, Y. et al. Conversational agent interventions for mental health problems: systematic review and meta-analysis of randomized controlled trials. J. Med. Internet Res. 25, e43862 (2023). source

      Not checked in the free run.

    • cites [20] Malik, T., Ambrose, A. J. & Sinha, C. Evaluating user feedback for an Artificial Intelligence–Enabled, cognitive behavioral therapy–based mental Health App (Wysa): qualitative thematic analysis. JMIR Hum. Factors 9, e35668 (2022). source

      Not checked in the free run.

    • cites [19] Ahmed, A. et al. Thematic analysis on user reviews for depression and anxiety Chatbot Apps: machine learning approach. JMIR Form. Res. 6, e27654 (2022). source

      Not checked in the free run.

    • cites [21] Ta, V. et al. User experiences of social support from companion chatbots in everyday contexts: thematic analysis. J. Med. Internet Res. 22, e16235 (2020). source

      Not checked in the free run.

    • cites [22] Haque, M. R. & Rubya, S. An overview of Chatbot-Based mobile mental health Apps: insights from App Description and user reviews. JMIR MHealth UHealth 11, e44838 (2023). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Generative AI's capabilities represent a significant opportunity for digital mental health, with media reports of increasing consumer usage, one meta-analysis finding generative AI chatbots more effective than rule-based ones at reducing psychological distress, and a pilot study showing promising results from ChatGPT usage in psychiatric inpatient care.

    “Generative AI’s capabilities represent a significant opportunity for digital mental health37, with media reports of increasing consumer usage38,39, one meta-analysis finding generative AI chatbots more effective than rule-based ones at reducing psychological distress40, and a pilot study showing promising results from ChatGPT usage in psychiatric inpatient care41.”
    Quote in the paperNot a substantial claim0 of 5 sources checked
    • cites [37] Torous, J. The Digital Mental Health Paradox: Is Now the Time to Unlock the Potential? Harv. Health Policy Rev. 23, (2023). source

      Not checked in the free run.

    • cites [38] Broderick, R. People are using AI for therapy, whether the tech is ready for it or not. Fast Company https://www.fastcompany.com/90836906/ai-therapy-koko-chatgpt (2023).

      Not checked in the free run.

    • cites [39] Robb, A. ‘He checks in on me more than my friends and family’: can AI therapists do better than the real thing? The Guardian https://www.theguardian.com/lifeandstyle/2024/mar/02/can-ai-chatbot-therapists-do-better-than-the-real-thing (2024).

      Not checked in the free run.

    • cites [40] Li, H., Zhang, R., Lee, Y.-C., Kraut, R. E. & Mohr, D. C. Systematic review and meta-analysis of AI-based conversational agents for promoting mental health and well-being. Npj Digit. Med. 6, 236 (2023). source

      Not checked in the free run.

    • cites [41] Melo, A., Silva, I. & Lopes, J. ChatGPT: A Pilot Study on a Promising Tool for Mental Health Support in Psychiatric Inpatient Care. Int. J. Psychiatr. Trainees https://doi.org/10.55922/001c.92367 (2024). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    User engagement remains a persistent challenge, with mixed user reviews, and studies indicating that 30 days after installation the proportion of users still active may be as low as 3%.

    “Moreover, user engagement remains a persistent challenge, with mixed user reviews11, and studies indicating that 30 days after installation the proportion of users still active may be as low as 3%12.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [11] Haque, M. R. & Rubya, S. ‘For an App Supposed to Make Its Users Feel Better, It Sure is a Joke’ - An Analysis of User Reviews of Mobile Mental Health Applications. Proc. ACM Hum. Comput. Interact. 6, 1–29 (2022). source

      Not checked in the free run.

    • cites [12] Baumel, A., Muench, F., Edan, S. & Kane, J. M. Objective user engagement with mental Health Apps: systematic search and panel-based usage analysis. J. Med. Internet Res. 21, e14567 (2019). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Unlike rule-based AI chatbots, generative AI chatbots like OpenAI's ChatGPT, Google's Gemini, and Inflection's Pi are trained on vast amounts of data, enabling them to understand and generate language with remarkable proficiency.

    “Unlike rule-based AI chatbots, generative AI chatbots like OpenAI’s ChatGPT, Google’s Gemini, and Inflection’s Pi are trained on vast amounts of data26, enabling them to understand and generate language with remarkable proficiency27.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [26] Vaswani, A. et al. Attention Is All You Need. Preprint at http://arxiv.org/abs/1706.03762 (2023).

      Not checked in the free run.

    • cites [27] Bubeck, S. et al. Sparks of Artificial General Intelligence: Early experiments with GPT-4. Preprint at http://arxiv.org/abs/2303.12712 (2023).

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    The evidence for the effectiveness of DMHIs has been limited, with a meta-analysis of randomised controlled trials (RCTs) finding only small effect sizes, potential publication bias, and a lack of active controls.

    “However, the evidence for the effectiveness of DMHIs has been limited, with a meta-analysis of randomised controlled trials (RCTs) finding only small effect sizes, potential publication bias, and a lack of active controls8–10.”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [8] Goldberg, S. B., Lam, S. U., Simonsson, O., Torous, J. & Sun, S. Mobile phone-based interventions for mental health: A systematic meta-review of 14 meta-analyses of randomized controlled trials.PLOS Digit. Health 1, e0000002 (2022). source

      Not checked in the free run.

    • cites [9] Groot, J. et al. The effectiveness of fully automated digital interventions in promoting mental well-being in the general population: systematic review and meta-analysis.JMIR Ment. Health 10, e44658 (2023). source

      Not checked in the free run.

    • cites [10] Garrido, S. et al. What works and what doesn’t work? A systematic review of digital mental health interventions for depression and anxiety in young people. Front. Psychiatry 10, 759 (2019). source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Rule-based chatbot apps have more promising user engagement, with positive app store ratings and qualitative studies finding that users appreciate the human-like interaction and social support.

    “Rule-based chatbot apps have more promising user engagement, with positive app store ratings19,20 and qualitative studies finding that users appreciate the human-like interaction and social support19–23.”
    Quote in the paperNot a substantial claim0 of 5 sources checked
    • cites [19] Ahmed, A. et al. Thematic analysis on user reviews for depression and anxiety Chatbot Apps: machine learning approach. JMIR Form. Res. 6, e27654 (2022). source

      Not checked in the free run.

    • cites [20] Malik, T., Ambrose, A. J. & Sinha, C. Evaluating user feedback for an Artificial Intelligence–Enabled, cognitive behavioral therapy–based mental Health App (Wysa): qualitative thematic analysis. JMIR Hum. Factors 9, e35668 (2022). source

      Not checked in the free run.

    • cites [21] Ta, V. et al. User experiences of social support from companion chatbots in everyday contexts: thematic analysis. J. Med. Internet Res. 22, e16235 (2020). source

      Not checked in the free run.

    • cites [22] Haque, M. R. & Rubya, S. An overview of Chatbot-Based mobile mental health Apps: insights from App Description and user reviews. JMIR MHealth UHealth 11, e44838 (2023). source

      Not checked in the free run.

    • cites [23] Kettle, L. & Lee, Y.-C. User experiences of well-being Chatbots. Hum. Factors J. Hum. Factors Ergon. Soc. 66, 1703–1723 (2024). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    Generative AI chatbots may offer mental health support that feels meaningful to users, but further research is needed on safety and effectiveness.

    “Generative AI chatbots may offer mental health support that feels meaningful to users, but further research is needed on safety and effectiveness.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.
    • Rests on an assumption: The sample of nineteen participants is sufficient to develop meaningful themes about experiences with generative AI chatbots.
    • Rests on an assumption: The themes developed from the interviews accurately represent the participants' experiences.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Digital mental health interventions (DMHIs) have emerged over the last decade as a promising potential response to the treatment gap, leveraging technology to deliver low-cost, effective, always-available and anonymous (and thus low-stigma) mental health treatment at scale.

    “Digital mental health interventions (DMHIs) have emerged over the last decade as a promising potential response to the treatment gap, leveraging technology to deliver low-cost, effective, always-available and anonymous (and thus low-stigma) mental health treatment at scale6.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [6] Torous, J., Benson, N. M., Myrick, K. & Eysenbach, G. Focusing on digital research priorities for advancing the access and quality of mental health. JMIR Ment. Health 10, e47898 (2023). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Generative AI brings new challenges, including potential risks of harm and questions of liability; trustworthiness issues such as the tendency to output incorrect or fabricated content (to 'hallucinate'), lack of predictability or interpretability, and inherent biases in training data; and the need to demonstrate clinical effectiveness.

    “However, this new technology also brings new challenges, including potential risks of harm and questions of liability42; trustworthiness issues such as the tendency to output incorrect or fabricated content (to “hallucinate”), lack of predictability or interpretability, and inherent biases in training data43; and the need to demonstrate clinical effectiveness44.”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [42] De Freitas, J. & Cohen, I. G. The health risks of generative AI-based wellness apps. Nat. Med. https://doi.org/10.1038/s41591-024-02943-6 (2024). source

      Not checked in the free run.

    • cites [43] Department for Science, Innovation and Technology, AI Safety Institute & Bengio, Y. International Scientific Report on the Safety of Advanced AI - Interim Report. UK Government https://www.gov.uk/government/publications/international-scientific-report-on-the-safety-of-advanced-ai (2024).

      Not checked in the free run.

    • cites [44] Chung, N. C., Dyer, G. & Brocki, L. Challenges of Large Language Models for Mental Health Counseling. Preprint at http://arxiv.org/abs/2311.13857 (2023).

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    User engagement with generative AI has been impressive, with ChatGPT's user base growing to 100 million weekly active users within a year of launch and an estimated half of the US population having used generative AI.

    “Furthermore, user engagement has been impressive, with ChatGPT’s user base growing to 100 million weekly active users within a year of launch34 and an estimated half of the US population having used generative AI35,36.”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [34] Malik, A. OpenAI’s ChatGPT now has 100 million weekly active users. TechCrunch AI https://techcrunch.com/2023/11/06/openais-chatgpt-now-has-100-million-weekly-active-users/ (2023).

      Not checked in the free run.

    • cites [35] Salesforce. Top Generative AI Statistics for 2024. Salesforce News & Insights https://www.salesforce.com/news/stories/generative-ai-statistics/ (2023).

      Not checked in the free run.

    • cites [36] Pandya. The Age of Generative AI: Over half of Americans have used generative AI and most believe it will help them be more creative. Adobe blog https://blog.adobe.com/en/publish/2024/04/22/age-generative-ai-over-half-americans-have-used-generative-ai-most-believe-will-help-them-be-more-creative (2024).

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Two popular chatbots, Woebot and Wysa, have been shown to improve users' depression symptoms, and build therapeutic alliances that appear comparable to those formed with human therapists.

    “For example, two popular chatbots, Woebot and Wysa, have been shown to improve users’ depression symptoms15,16, and build therapeutic alliances that appear comparable to those formed with human therapists17,18.”
    Quote in the paperNot a substantial claim0 of 4 sources checked
    • cites [15] Fitzpatrick, K. K., Darcy, A. & Vierhile, M. Delivering cognitive behavior therapy to young adults with symptoms of depression and anxiety using a fully automated conversational agent (Woebot): a randomized controlled trial. JMIR Ment. Health 4, e19 (2017). source

      Not checked in the free run.

    • cites [16] Inkster, B., Sarda, S. & Subramanian, V. An empathy-driven, conversational artificial intelligence agent (Wysa) for digital mental well-being: real-world data evaluation mixed-methods study. JMIR MHealth UHealth 6, e12106 (2018). source

      Not checked in the free run.

    • cites [17] Beatty, C., Malik, T., Meheli, S. & Sinha, C. Evaluating the therapeutic alliance with a free-text CBT Conversational Agent (Wysa): a mixed-methods study. Front. Digit. Health 4, 847991 (2022). source

      Not checked in the free run.

    • cites [18] Darcy, A., Daniels, J., Salinger, D., Wicks, P. & Robinson, A. Evidence of human-level bonds established with a digital conversational agent: cross-sectional, retrospective observational study. JMIR Form. Res. 5, e27868 (2021). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationnone given

    Some themes echoed prior research on rule-based chatbots, while others seemed novel to generative AI.

    “Some themes echoed prior research on rule-based chatbots, while others seemed novel to generative AI.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.
    • Rests on an assumption: The sample of nineteen participants is sufficient to develop meaningful themes about experiences with generative AI chatbots.
    • Rests on an assumption: The themes developed from the interviews accurately represent the participants' experiences.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Generative AI models are increasingly achieving or surpassing human performance benchmarks in various domains, including medical diagnostic dialogue, persuasive communication, theory of mind, making people feel heard, responding to relationship issues and helping people reframe negative situations to reduce negative emotions.

    “These models are increasingly achieving or surpassing human performance benchmarks in various domains, including medical diagnostic dialogue28, persuasive communication29, theory of mind30, making people feel heard31, responding to relationship issues32 and helping people reframe negative situations to reduce negative emotions33.”
    Quote in the paperNot a substantial claim0 of 6 sources checked
    • cites [28] Tu, T. et al. Towards Conversational Diagnostic AI. Preprint at http://arxiv.org/abs/2401.05654 (2024).

      Not checked in the free run.

    • cites [29] Salvi, F., Ribeiro, M. H., Gallotti, R. & West, R. On the conversational persuasiveness of large language models: a randomized controlled trial. Preprint at http://arxiv.org/abs/2403.14380 (2024). source

      Not checked in the free run.

    • cites [30] Strachan, J. W. A. et al. Testing theory of mind in large language models and humans. Nat. Hum. Behav. https://doi.org/10.1038/s41562-024-01882-z (2024). source

      Not checked in the free run.

    • cites [31] Yin, Y., Jia, N. & Wakslak, C. J. AI can help people feel heard, but an AI label diminishes this impact. Proc. Natl Acad. Sci. USA 121, e2319112121 (2024). source

      Not checked in the free run.

    • cites [32] Young, J. et al. The Role of AI in Peer Support for Young People: A Study of Preferences for Human- and AI-Generated Responses. in Proc. CHI Conference on Human Factors in Computing Systems 1–18 (ACM, Honolulu HI USA, 2024). https://doi.org/10.1145/3613904.3642574. source

      Not checked in the free run.

    • cites [33] Li, J. Z., Herderich, A. & Goldenberg, A. Skill but not Effort Drive GPT Overperformance over Humans in Cognitive Reframing of Negative Scenarios. Preprint at https://doi.org/10.31234/osf.io/fzvd8 (2024). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Rule-based AI chatbots show promise to address some of these limitations, by simulating human conversation using predefined scripts and algorithms such as decision trees, to deliver the benefits of DMHIs in a more dynamic and interactive way.

    “Rule-based AI chatbots show promise to address some of these limitations, by simulating human conversation using predefined scripts and algorithms such as decision trees, to deliver the benefits of DMHIs in a more dynamic and interactive way13,14.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [13] Vaidyam, A. N., Wisniewski, H., Halamka, J. D., Kashavan, M. S. & Torous, J. B. Chatbots and conversational agents in mental health: a review of the psychiatric landscape. Can. J. Psychiatry 64, 456–464 (2019). source

      Not checked in the free run.

    • cites [14] Lim, S. M., Shiau, C. W. C., Cheng, L. J. & Lau, Y. Chatbot-delivered psychotherapy for adults with depressive and anxiety symptoms: a systematic review and meta-regression. Behav. Ther. 53, 334–347 (2022). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

    Interpretationown data

    The absence of reports of harmful behaviours should not be considered evidence of absence, but more research may be warranted to assess whether the risks have changed with recent technological improvements, or whether these issues are simply rarer, requiring larger sample sizes to uncover.

    “This should not be considered evidence of absence, but more research may be warranted to assess whether the risks have changed with recent technological improvements, or whether these issues are simply rarer, requiring larger sample sizes to uncover.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Premisecitation

    Inflection, the company behind the Pi chatbot, pivoted in March 2024 from providing consumer emotional support services towards enterprise AI services, due to a lack of a business model, despite USD 1.5 billion of investment.

    “Inflection, the company behind the Pi chatbot used by most of the participants in our study, pivoted in March 2024 from providing consumer emotional support services towards enterprise AI services, due to a lack of a business model, and despite USD 1.5 billion of investment60.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [60] Ghaffary, S. Inflection AI Plans Pivot After Microsoft Hirings. Bloomberg https://www.bloomberg.com/news/articles/2024-03-19/inflection-ai-plans-pivot-after-most-employees-go-to-microsoft (2024).

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Interpretationown data

    Generative AI chatbot usage is constantly evolving as the underlying technology improves.

    “These simple questions may not yield clear answers, as our study shows that generative AI chatbot usage is diverse, complex and personalised, and moreover, constantly evolving as the underlying technology improves.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    Current literature tends to advocate for a cautious approach, in which near-term clinical generative AI applications are limited to implementations of evidence-based therapies (such as CBT), with a clinician in the loop, and models constrained to scripted responses as far as possible.

    “Current literature tends to advocate for a cautious approach, in which near-term clinical generative AI applications are limited to implementations of evidence-based therapies (such as CBT)52, with a clinician in the loop52, and models constrained to scripted responses as far as possible42.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [52] Stade, E. C. et al. Large language models could change the future of behavioral healthcare: a proposal for responsible development and evaluation. Npj Ment. Health Res. 3, 12 (2024). source

      Not checked in the free run.

    • cites [42] De Freitas, J. & Cohen, I. G. The health risks of generative AI-based wellness apps. Nat. Med. https://doi.org/10.1038/s41591-024-02943-6 (2024). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    Other themes appear to be more novel, such as the level of joy experienced, the sense of being deeply understood, the breadth and quality of advice, and the ability to work on mental health in flexible and creative ways, such as through role-play, imagery and fiction.

    “Other themes appear to be more novel, such as the level of joy experienced, the sense of being deeply understood, the breadth and quality of advice, and the ability to work on mental health in flexible and creative ways, such as through role-play, imagery and fiction.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.
    • Rests on an assumption: The sample of nineteen participants is sufficient to develop meaningful themes about experiences with generative AI chatbots.
    • Rests on an assumption: The themes developed from the interviews accurately represent the participants' experiences.

  • May combine two claims.

    Interpretationreasoning

    Our findings point to similarities and differences in how generative AI and rule-based chatbots are experienced.

    “Our findings point to similarities and differences in how generative AI and rule-based chatbots are experienced.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.
    • Rests on an assumption: The sample of nineteen participants is sufficient to develop meaningful themes about experiences with generative AI chatbots.
    • Rests on an assumption: The themes developed from the interviews accurately represent the participants' experiences.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Premisecitation

    Given the novelty of generative AI and the nascent state of the field, qualitative research is an important starting point to generate rich foundational insights into individuals' subjective experiences, which can be overlooked in quantitative studies.

    “Given the novelty of generative AI and the nascent state of the field, qualitative research is an important starting point to generate rich foundational insights into individuals’ subjective experiences, which can be overlooked in quantitative studies47.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [47] Braun, V. & Clarke, V. Using thematic analysis in psychology. Qual. Res. Psychol. 3, 77–101 (2006). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Guardrails like these are commonly implemented in consumer generative AI applications.

    “Guardrails like these are commonly implemented in consumer generative AI applications50.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [50] Dong, Y. et al. Building Guardrails for Large Language Models. Preprint at http://arxiv.org/abs/2402.01822 (2024).

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Qualitative studies published so far include thematic analyses of user forum comments on both generative AI and rule-based DMHIs, student survey responses on companion-focused generative AI chatbots, and semi-structured interviews with hospital outpatients who were asked to try ChatGPT for mental health support.

    “Qualitative studies published so far include thematic analyses of user forum comments on both generative AI and rule-based DMHIs23, student survey responses on companion-focused generative AI chatbots48, and semi-structured interviews with hospital outpatients who were asked to try ChatGPT for mental health support49.”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [23] Kettle, L. & Lee, Y.-C. User experiences of well-being Chatbots. Hum. Factors J. Hum. Factors Ergon. Soc. 66, 1703–1723 (2024). source

      Not checked in the free run.

    • cites [48] Maples, B., Cerit, M., Vishwanath, A. & Pea, R. Loneliness and suicide mitigation for students using GPT3-enabled chatbots. Npj Ment. Health Res. 3, 4 (2024). source

      Not checked in the free run.

    • cites [49] Alanezi, F. Assessing the effectiveness of ChatGPT in delivering mental health support: a qualitative study. J. Multidiscip. Healthc. 17, 461–471 (2024). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    Leading the therapeutic process would require chatbots to remember the conversation and build an internal model of their user, something that a majority of users currently miss.

    “Leading the therapeutic process would require chatbots to remember the conversation and build an internal model of their user, something that a majority of users currently miss –”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Premisecitation

    Given the unpredictable 'black box' nature of generative AI, and the existence of at least one tragic example of early generative AI chatbots supporting users in dying by suicide, current literature advocates that when users display signs of crisis, models revert to scripted responses that signpost towards human support.

    “Given the unpredictable “black box” nature of generative AI43, and the existence of at least one tragic example of early generative AI chatbots supporting users in dying by suicide55, current literature advocates that when users display signs of crisis, models revert to scripted responses that signpost towards human support42,51.”
    Quote in the paperSubstantial claim0 of 4 sources checked
    • cites [43] Department for Science, Innovation and Technology, AI Safety Institute & Bengio, Y. International Scientific Report on the Safety of Advanced AI - Interim Report. UK Government https://www.gov.uk/government/publications/international-scientific-report-on-the-safety-of-advanced-ai (2024).

      Not checked in the free run.

    • cites [55] Pierre-François, L. Without these conversations with the Eliza chatbot, my husband would still be here. La Libre https://www.lalibre.be/belgique/societe/2023/03/28/sans-ces-conversations-avec-le-chatbot-eliza-mon-mari-serait-toujours-la-LVSLWPC5WRDX7J2RCHNWPDST24/ (2023).

      Not checked in the free run.

    • cites [42] De Freitas, J. & Cohen, I. G. The health risks of generative AI-based wellness apps. Nat. Med. https://doi.org/10.1038/s41591-024-02943-6 (2024). source

      Not checked in the free run.

    • cites [51] Heston, T. F. Safety of Large Language Models in Addressing Depression. Cureus https://doi.org/10.7759/cureus.50729 (2023). source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Recent research shows that generative AI can help halt suicidal ideation, and that young people show a preference for generative AI support responses over those from peers, adult mentors and therapists - but not on topics that invoke the AI's safety guardrails.

    “This resonates with recent research showing that generative AI can help halt suicidal ideation48, and that young people show a preference for generative AI support responses over those from peers, adult mentors and therapists – but not on topics that invoke the AI’s safety guardrails32.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [48] Maples, B., Cerit, M., Vishwanath, A. & Pea, R. Loneliness and suicide mitigation for students using GPT3-enabled chatbots. Npj Ment. Health Res. 3, 4 (2024). source

      Not checked in the free run.

    • cites [32] Young, J. et al. The Role of AI in Peer Support for Young People: A Study of Preferences for Human- and AI-Generated Responses. in Proc. CHI Conference on Human Factors in Computing Systems 1–18 (ACM, Honolulu HI USA, 2024). https://doi.org/10.1145/3613904.3642574. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    Providing the safest response to those in crisis may require a more nuanced, balanced and sophisticated approach, based on a more complete understanding of capabilities and risks.

    “Therefore, providing the safest response to those in crisis may require a more nuanced, balanced and sophisticated approach, based on a more complete understanding of capabilities and risks.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • May combine two claims.

    Interpretationown data

    The current approach to safety guardrails may be oversimplified in two ways: (1) by underestimating the capabilities of generative AI to respond to crises, and (2) by limiting those capabilities at the times that matter most.

    “But this approach may be oversimplified in two ways: (1) by underestimating the capabilities of generative AI to respond to crises, and (2) by limiting those capabilities at the times that matter most.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The participants' self-reported experiences and impacts are accurate and can be taken at face value.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductown data

    The consent form was signed by 35 individuals, of which 19 subsequently booked and attended an interview.

    “The consent form was signed by 35 individuals, of which 19 subsequently booked and attended an interview.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    There is a degree of subjectivity in how themes are developed in reflexive thematic analysis, especially when conducted by a sole researcher.

    “As with all reflexive thematic analysis, there is a degree of subjectivity in how themes are developed, especially when conducted by a sole researcher (SS).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductcitation

    The topic guide was drafted following the framework from Kallio et al., reviewed with the second author, and piloted with a research collaborator before starting the interviews.

    “Following the framework from Kallio et al.67, the first author (SS) drafted a topic guide, inspired by existing qualitative research in this area, reviewed with the second author (AC), and piloted with a research collaborator before starting the interviews”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [67] Kallio, H., Pietilä, A., Johnson, M. & Kangasniemi, M. Systematic methodological review: developing a framework for a qualitative semi‐structured interview guide. J. Adv. Nurs. 72, 2954–2965 (2016). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductreasoning

    AC reviewed the codes and themes, helping to ensure rigour and validity.

    “with AC’s reviews of codes and themes helping to ensure rigour and validity”
    • Also stated in: § Data analysis · plain
      “resulting in around 600 codes, which were reviewed by AC”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductcitation

    The study followed Braun & Clarke's reflexive thematic analysis approach, taking an inductive, data-driven approach without reference to any preconceived theory or framework.

    “We followed Braun & Clarke’s reflexive thematic analysis approach to code the transcripts and develop themes, taking an inductive approach, i.e., in an open-ended and data-driven way, without reference to any preconceived theory or framework47,68,69.”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [47] Braun, V. & Clarke, V. Using thematic analysis in psychology. Qual. Res. Psychol. 3, 77–101 (2006). source

      Not checked in the free run.

    • cites [68] Byrne, D. A worked example of Braun and Clarke’s approach to reflexive thematic analysis. Qual. Quant. 56, 1391–1412 (2022). source

      Not checked in the free run.

    • cites [69] Braun, V. & Clarke, V. One size fits all? What counts as quality practice in (reflexive) thematic analysis? Qual. Res. Psychol. 18, 328–352 (2021). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductown data

    The resulting codes, subthemes and themes were shared with two participants who had requested their transcripts, with an invitation to feedback; no corrections were provided.

    “The resulting codes, subthemes and themes were shared with two participants who had requested their transcripts, with an invitation to feedback if anything appeared misrepresented; no corrections were provided.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Semi-structured interviews are a well-established approach to enable participants to express diverse perceptions and focus on topics most meaningful to them, particularly for complex or emotionally sensitive topics.

    “We collected data using semi-structured interviews, as a well-established approach to enable participants to express diverse perceptions and focus on topics most meaningful to them, in particular for complex or emotionally sensitive topics that they may not be used to discussing with others67.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [67] Kallio, H., Pietilä, A., Johnson, M. & Kangasniemi, M. Systematic methodological review: developing a framework for a qualitative semi‐structured interview guide. J. Adv. Nurs. 72, 2954–2965 (2016). source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    Participants were required to have had at least three separate conversations with an LLM-based generative AI chatbot on mental health and wellbeing topics, each lasting at least 20 minutes, to ensure enough breadth and depth of usage.

    “Participants were required to have had at least three separate conversations with an LLM-based generative AI chatbot on mental health and wellbeing topics, each lasting at least 20 minutes.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    AC, an expert in qualitative methods, reviewed and quality-checked the video of the first interview.

    “AC, an expert in qualitative methods, reviewed and quality-checked the video of the first interview.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    The study may miss important experiences from individuals where the mental health treatment gap is most urgent, and from individuals for whom the technology did not work.

    “This study may miss important experiences from individuals where the mental health treatment gap is most urgent, and from individuals for whom the technology did not work.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductown data

    Most participants lived in high-income countries, were tech-savvy and well-educated, and focused on milder forms of mental health conditions.

    “Most of our participants lived in high-income countries, were tech-savvy and well-educated, and focused on milder forms of mental health conditions”
    Quote in the paperSubstantial claim
65 of 65

  • The paper relies on this without support.

    The participants' use of generative AI chatbots for mental health occurred in unprompted, unguided real-world settings.

    “in unprompted, unguided real-world settings”

  • The paper relies on this without support.

    The participants' self-reported experiences and impacts are accurate and can be taken at face value.

    “We interviewed nineteen individuals about their experiences using generative AI chatbots for mental health.”

  • The paper relies on this without support.

    The sample of nineteen participants is sufficient to develop meaningful themes about experiences with generative AI chatbots.

    “Nineteen participants (12 male, 7 female) were recruited to the study.”

  • The paper relies on this without support.

    The themes developed from the interviews accurately represent the participants' experiences.

    “Four overarching themes were developed”

  • The paper relies on this without support.

    There is a lack of information on user populations of generative AI for mental health.

    “Given the emerging nature of generative AI use for mental health and the attendant lack of information on user populations, convenience sampling was employed as a pragmatic approach to maximise the likelihood of finding sufficient real-world, unprompted participants.”
5 of 5

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?

  • The mental health support potential of LLM-CBs has been noted by both patients and practitioners.

    “The mental health support potential of LLM-CBs has been noted by both patients149,152,153 and practitioners153–155”

    From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [152]

    The Discussion reports that participants using generative AI chatbots described meaningful mental health support, supporting the patient side of the claim. It does not report practitioners noting that potential: its clinician-focused passage recommends that clinicians learn about the tools and says a survey found therapists had reservations. The indexed passages have section labels but no page numbers. (LLM, conf=0.88)

    Experimental — accuracy not yet measured

  • This capacity may strengthen user engagement, therapeutic alliance, and the sense of being understood.

    “This capacity may strengthen user engagement [20], therapeutic alliance [21], and the sense of being understood [22].”

    From Generative AI Mental Health Chatbots as Therapeutic Tools: Systematic Review and Meta-Analysis of Their Role in Reducing Mental Health Issues (2025) · cites this paper as [21]

  • Some users found chatbots lacked empathy, could not lead the conversation, and struggled to remember things.

    “Some users still found that chatbots lacked empathy, could not lead the conversation, and struggled to remember things.”

    From Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers. (2025) · cites this paper as [125]

The citation lists haven’t been fetched yet

References 57 · Cited by 241

Authors
Siddals, S., Torous, J., & Coxon, A.
Affiliations
  • King's College London (GB)
  • Beth Israel Deaconess Medical Center (US)
Corresponding author
Siddals, Steven
ORCID
Venue
npj Mental Health Research
Publisher
Springer Science and Business Media LLC
Published
2024-10-27
Volume
3
Issue
1
Pages
48-48
ISSN
2731-4251 (ISSN-L)
DOI
10.1038/s44184-024-00097-4
PMC
PMC11514308
OpenAlex
W4403843885
Semantic Scholar
c840f690f0e1ab5ed6c23d79140d0709e13c4d09
Cited by
241
References
57
Citations by year
2026: 155 · 2025: 85 · 2024: 1
FWCI
95.57
Type
journal article
Language
en
Open access
Licence
Locations
  • npj Mental Health Research (publishedVersion, cc-by)
  • Research Square (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
  • Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
Keywords
generative AI chatbots, mental health, user experience, ChatGPT, trauma recovery, loss, safety guardrails, therapeutic process

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.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-07.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-07.

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