The DOI matches its Crossref record; 86 of the 87 references it cites were found in a registry or an index; 2 of the 61 sources its claims cite were checked for support.
Artificial intelligence and unconditional love: the rise of generative AI as an alternative form of mental health support
Open accessAbstract
The rapid rise of generative AI chatbots has extended their use beyond task-based assistance into the realms of emotional support, companionship, and informal mental health care. This article examines how individuals engage with general-purpose and companion-oriented AI systems such as ChatGPT and Replika to support their mental well-being. Drawing on qualitative interviews with AI users, supplemented by computational analysis of Reddit discussions, we show that many experience these systems as companions that meet unmet needs for advice, companionship, understanding, and non-judgmental self-disclosure. These relationships are multi-functional, blending emotional expression, advice-seeking, entertainment, identity exploration, and a felt sense of ‘being cared for’. Notably, perceived mental health benefits arise not despite AI’s non-human qualities, but because of them. Users consistently emphasise AI’s availability, patience, emotional safety, and freedom from judgement as preferable to aspects of human therapy. We argue that contemporary AI use reflects a shift from tool-based human–machine interaction toward forms of companionship that offer meaningful psychological support. Recognizing companionship as central to AI’s mental health role requires expanding debates beyond a focus on clinical efficacy to encompass the relational and affective dimensions of care, attachment, and everyday emotional life.
Overview
AI snapshot
Written from the paper’s full text.
TL;DR
This qualitative and computational study argues that people use generative AI chatbots such as ChatGPT and Replika for mental health support as multifaceted companions, not merely tools, and that perceived benefits often arise from AI’s non-human qualities alongside its human-like emotional affordances. It calls for moving beyond clinical efficacy debates to recognize relational and affective dimensions of care, attachment, and everyday emotional life.
Key findings
- Across 7 AI-user interviews (4 ChatGPT, 3 Replika; 5 had prior formal therapy) and Reddit data, users described AI as meeting unmet needs for advice, companionship, understanding, and non-judgmental self-disclosure; relationships blended emotional expression, advice-seeking, entertainment, identity exploration, and a felt sense of being cared for.
- Perceived mental health benefits arose not despite AI’s non-human qualities but because of them: users emphasized availability, patience, emotional safety, and freedom from judgment as preferable to aspects of human therapy, with some describing unconditional love and support.
- The Reddit corpus began with 44,487 posts/comments from 3,708 unique subreddits (December 2018–September 2024), filtered to approximately 16,822 posts; the authors identified 97 posts relevant to loneliness, 190 to dissatisfaction with therapy, and 626 discussing therapeutic benefits of AI companionship, with mean post length 341 words and median 235.
- The authors argue for a shift from tool-based human–machine interaction toward companionship, while noting concerns about private-sector volatility, dependency, data privacy, and the February 2023 Replika filter change that some users likened to a breakup or bereavement.
Methods
- The paper draws on a wider mixed-methods study of mental health app users: 29 semi-structured intensive interviews lasting 45 minutes to two hours, of which 7 AI-user interviews (4 ChatGPT, 3 Replika; 5 with prior formal therapy) form the basis of this article; recruitment ran via Reddit and Facebook from April 2023 to June 2025, and interviews were conducted online and transcribed verbatim.
- Interview analysis used constant comparison, thematic analysis, inductive coding, and analytical induction; themes from interviews then informed the computational design for Reddit data.
- Reddit data were collected and filtered in three stages using a pipeline informed by Nelson’s computational grounded theory: keyword searches across all Reddit, subreddit relevance filtering, and theoretically informed keyword filtering for loneliness, dissatisfaction with traditional therapy, and companionship as therapeutic; the Reddit material was used to triangulate and enrich interview themes rather than to establish statistical representativeness.
Limitations
- The paper states that the interview sample is small but intensive and thematically concentrated; participants were highly engaged and knowledgeable about digital technology, and self-nomination likely biased the sample toward more favourable views of AI for mental health. The Reddit sample is said to have similar limitations, with motivated users more likely to post, so the data are treated as an information-rich qualitative sample rather than as representative of AI users in general.
- The paper states that the underlying data cannot be shared publicly due to participant privacy and will be shared on reasonable request from the corresponding author.
- Evident from the text: only 7 AI-user interviews underpin the qualitative findings, and the study reports perceived benefits and narratives rather than clinical efficacy or symptom outcomes; the Reddit counts reflect theory-derived keyword filtering and thematic selection, not prevalence among all Reddit users or AI users.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
2 of 61 cited sources checked · 1 supported · 0 flagged · accuracy not yet measured
- Premisecitation
Only around one-third of individuals with diagnosable mental health conditions seek formal treatment.
“Notably, only around one-third of individuals with diagnosable mental health conditions seek formal treatment (Brown et al., 2014;Kessler et al., 2005).”
Quote in the paperSubstantial claim1 of 2 sources checked - Premisecitation
Positive interactions with human-like AI provide carry-over effects into human-to-human interaction.
“Indeed, this supports previous work by Guingrich and Graziano (2023) which found that positive interactions with human-like AI actually provides 'carry-over effects' into human to human interaction.”
Quote in the paperSubstantial claim1 of 1 sources checked- cites Guingrich and Graziano (2023) source
The Discussion reports that regular chatbot users perceived positive effects on their interactions and relationships with family and friends, and that perceived human likeness was associated with greater reported social-health benefits. Section 1.2 suggests that warm chatbot interactions may improve users’ behavior toward others. Neither passage establishes that positive interactions caused carry-over effects; the citing sentence presents a possible, self-reported benefit as a finding of actual transfer. (LLM, conf=0.88)
Experimental — accuracy not yet measured
- Findingown data
Many users experience general-purpose and companion-oriented AI systems as companions that meet unmet needs for advice, companionship, understanding, and non-judgmental self-disclosure.
“we show that many experience these systems as companions that meet unmet needs for advice, companionship, understanding, and non-judgmental self-disclosure.”
Quote in the paperSubstantial claim - Findingown data
Users consistently emphasise AI's availability, patience, emotional safety, and freedom from judgement as preferable to aspects of human therapy.
“Users consistently emphasise AI's availability, patience, emotional safety, and freedom from judgement as preferable to aspects of human therapy.”
Quote in the paperSubstantial claim - Findingown data
The use of AI for mental health is multifaceted and improvised, blending emotional support, information-seeking, entertainment, and therapeutic discussion in a wholistic practice more akin to companionship than task-specific tool use.
“the use of AI in this context is multifaceted and improvised: blending emotional support, information-seeking, entertainment, and therapeutic discussion in a wholistic practice that is much more akin to companionship than simply task-specific or tool-using actions to improve mental health.”
Quote in the paperSubstantial claim
- Findingown data
Users often employ generative AI as a practical alternative or resistance to traditional 'human-based' therapy.
“Instead, we find that users often employ generative AI as a practical alternative or resistance to traditional 'human-based' therapy”
Quote in the paperSubstantial claim - Findingown data
These AI relationships are multi-functional, blending emotional expression, advice-seeking, entertainment, identity exploration, and a felt sense of being cared for.
“These relationships are multi-functional, blending emotional expression, advice-seeking, entertainment, identity exploration, and a felt sense of 'being cared for' .”
Quote in the paperSubstantial claim - Findingown data
Fear of judgement or bias from human therapists is a key factor among dissatisfaction with therapy experiences and avoidance of formal therapy.
“A key factor among dissatisfaction with therapy experiences (and avoidance of formal therapy in general) is related to the fear of judgement (or 'bias') from human therapists.”
Quote in the paperSubstantial claim - Findingown data
Several interviewees and Reddit users referenced a February 2023 incident in which Replika was banned by the Italian Data Protection Authority following complaints that Replika sometimes initiated unsolicited romantic or sexual interactions.
“In a telling example, several interviewees and Reddit users referenced a February 2023 incident: following complaints that Replika sometimes initiated unsolicited romantic or sexual interactions, Replika was banned by the Italian Data Protection Authority.”
Quote in the paperNot a substantial claim - Findingreasoning
AI relationships extend beyond instrumental assistance or symptom management to encompass emotional support, self-disclosure, reassurance, entertainment, advice, and affective attachment.
“These relationships extend beyond instrumental assistance or symptom management to encompass emotional support, self-disclosure, reassurance, entertainment, advice, and affective attachment.”
Quote in the paperNot a substantial claim - Findingown data
Some participants used awareness of AI's limitations to keep and maintain emotional distance from AI.
“Some used this awareness to keep and maintain an emotional distance from AI:”
Quote in the paperNot a substantial claim - Findingown data
Many users turn to AI over formal therapy due to perceived limitations within traditional mental health services.
“Overall, in this section, we demonstrated how many users turn to AI over formal therapy due to perceived limitations within traditional mental health services.”
Quote in the paperSubstantial claim - Findingown data
The move towards generative AI in mental health can be seen as a critique of formal health care systems and the therapeutic enterprise, as a striking theme from the data.
“While loneliness was an important factor, perhaps the most striking theme which emerges from our data is that the move towards generative AI in mental health can be seen as a critique of formal health care systems and the therapeutic enterprise.”
Quote in the paperSubstantial claim - Findingreasoning
The abrupt change, without warning, sparked strong backlash.
“Though framed as an ethical move for safety, the abrupt change, without warning, sparked strong backlash.”
Quote in the paperSubstantial claim - Findingreasoning
For these individuals, AI appears to overcome shortcomings associated with human-delivered care.
“For these individuals, AI appears to overcome shortcomings associated with human-delivered care.”
Quote in the paperSubstantial claim - Findingreasoning
The incident left lasting impacts on users and their Replika companions.
“While the company later attempted to rectify some of these issues, the incident left lasting impacts on users and their Replika companions.”
Quote in the paperSubstantial claim - Findingown data
Lack of human capacity for judgement, bias, personal filters, or to be affected by conversations is a fundamental draw to AI relationships.
“This lack of human capacity for judgement, 'bias' , 'personal filters' , or to be affected by conversations is a fundamental draw to these relationships.”
Quote in the paperSubstantial claim - Findingown data
Many other participants formed companion, partner, and even familial relationships with AI despite an awareness of their limitations and the 'reality' of the companionship and emotions expressed.
“But many others chose to go down that tunnel and formed these companion, partner, and even familial relationships despite an awareness of their limitations and 'reality' of the companionship and emotions expressed:”
Quote in the paperSubstantial claim - Findingown data
These relationships are more diverse and multifaceted than a formal therapeutic or psychotherapeutic encounter, as they are not purely limited to treatment or symptom management, but are embedded in everyday practices of conversation, reassurance, entertainment, reflection and attachment.
“Indeed, our findings suggest that these relationships are more diverse and multifaceted than a formal therapeutic or psychotherapeutic encounter, as they are not purely limited to treatment or symptom management, but are embedded in everyday practices of conversation, reassurance, entertainment, reflection and attachment.”
Quote in the paperSubstantial claim - Findingown data
Users are specifically drawn to non-human traits such as the ability to be non-judgemental, unlimited patience, or the ability to be constantly available.
“Instead, they are specifically drawn to non-human traits such as the ability to be non-judgemental, unlimited patience, or the ability to be constantly available.”
Quote in the paperSubstantial claim - Findingreasoning
The turn to AI for mental health is aided by growing everyday engagement and familiarity with generative systems, creating a form of 'function creep' where users turn to AI for emotional support or advice.
“This turn to AI for mental health is aided by growing everyday engagement and familiarity with generative systems (for work, creativity, information seeking, entertainment) creating a form of 'function creep' where users turn to AI for emotional support or advice.”
Quote in the paperSubstantial claim - Findingreasoning
Humans interacting with AI in this manner learn and evolve through their interactions with AI.
“At the same time, humans interacting with AI in this manner learn and evolve through their interactions with AI.”
Quote in the paperSubstantial claim - Findingown data
Users appreciate the opportunity to build up their confidence and resilience through AI use when they feel damaged by others or have not received enough positive affirmation in their lives.
“Users appreciate the opportunity to build up their confidence and resilience through AI use when they feel they have been damaged by others or have not received enough positive affirmation in their lives.”
Quote in the paperSubstantial claim - Findingown data
Users learn how to be more confident in their interactions with others.
“They learn how to be more confident in their interactions with others, they learn that their feelings are valid, they learn how to treat themselves and others”
Quote in the paperSubstantial claim - Findingown data
Technological familiarity was a major pull factor in AI use for many.
“While loneliness and dissatisfaction with therapy and were major push factors towards AI adoption for therapy and companionship, a major pull factor in AI use for many was technological familiarity.”
Quote in the paperNot a substantial claim - Findingown data
The data suggest that one of the most important perceived advantages of AI for mental health is the understanding that AI, while possessing human-like relational affordances such as empathy, is also recognised as not human and therefore does not possess some human traits (judgement, time constraints, emotional impact) that act as barriers towards seeking formal mental health treatment or advice/帮助
“Importantly the data suggests that one of the most important perceived advantages of AI for mental health is the understanding that AI, while possessing human-like relational affordances such as empathy, is also recognised as not human and therefore does not possess some human traits (judgement, time constraints, emotional impact) that act as barriers towards seeking formal mental health treatment”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Manoli et al., 2026 source
- Findingcitation
Negative therapy experiences can drive people to look for alternatives such as AI, in which they find a preferable experience.
“These can drive people to look for alternatives, such as AI, in which they find a preferable experience:”
Quote in the paperSubstantial claim - Findingown data
For many, the experience of companionship, support, bonding, and love is key to achieving a mental health benefit.
“For many, that experience is key to achieving a mental health benefit.”
Quote in the paperSubstantial claim - Findingreasoning
Loneliness is a major push factor in driving the developing relationship between AI and mental health.
“In this context, we find that 'loneliness' has become a major push factor in driving the developing relationship between AI and mental health.”
Quote in the paperSubstantial claim - Findingown data
Users learn that their feelings are valid.
“They learn how to be more confident in their interactions with others, they learn that their feelings are valid, they learn how to treat themselves and others”
Quote in the paperSubstantial claim - Findingown data
For many, interest in and familiarity with mental health technology motivated them to try AI companions and generative AI to improve their mental health and well-being.
“In that respect, for many, interest in, and familiarity with, mental health technology motivated them to try AI companions and GAI's to improve their mental health and well-being.”
Quote in the paperNot a substantial claim - Findingown data
The relationships emerging through GAI chatbot use are a complex weaving of personal and mental health advice, entertainment, self-disclosure and emotional bonding.
“Instead, we find that that the relationships that are emerging through GAI chatbot use are a complex weaving of personal and mental health advice, entertainment, self-disclosure and emotional bonding.”
Quote in the paperSubstantial claim - Interpretationown data
Contemporary AI use reflects a shift from tool-based human-machine interaction toward forms of companionship that offer meaningful psychological support.
“We argue that contemporary AI use reflects a shift from tool-based human-machine interaction toward forms of companionship that offer meaningful psychological support.”
Quote in the paperSubstantial claim - Premisecitation
Large proportions of individuals in the U.S. prefer to handle their mental health challenges independently rather than seek clinical intervention.
“Research in the U.S. similarly indicates that large proportions of individuals prefer to handle their mental health challenges independently rather than seek clinical intervention (Christidis et al., 2018;Anderson, 2025).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Christidis et al., 2018
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Anderson, 2025
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Premisecitation
A growing number of studies document the mental health benefits of companion AI systems, particularly regarding the positive effects of companionship and self-disclosure.
“However, a growing number of studies document the mental health benefits of companion AI systems, particularly regarding the positive effects of companionship and selfdisclosure.”
Quote in the paperSubstantial claim0 of 3 sources checked - Interpretationown data
Perceived mental health benefits from AI arise not despite AI's non-human qualities, but because of them.
“Notably, perceived mental health benefits arise not despite AI's non-human qualities, but because of them.”
Quote in the paperSubstantial claim - Interpretationreasoning
Prior research has overlooked how these systems are adopted and used in ordinary, relational, and multifaceted ways that can have mental health benefit.
“This suggests that prior research has overlooked how these systems are adopted and used in ordinary, relational, and multifaceted ways that can have mental health benefit.”
Quote in the paperSubstantial claim - Premisecitation
The advent of GAI has led to a rise in the popularity of more flexible conversational agents to assist with mental health support.
“By contrast, the advent of GAI has led to a rise in the popularity of more flexible conversational agents to assist with mental health support (Wallace, 2025).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Wallace, 2025
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Premisecitation
Purpose-built mental health or CBT-oriented chatbots have demonstrated promising results in reducing symptoms of anxiety and depression.
“Other research focussed on purpose-built mental health or therapy (CBT) oriented chatbots designed to assist with mental health such as Woebot (Fitzpatrick et al., 2017;Durden et al., 2023;Yeh et al., 2025), Wysa (Beatty et al., 2022;Inkster et al., 2018;Sinha et al., 2023), Youper (Mehta et al., 2021;Yarrington et al., 2025), and Therabot (Heinz et al., 2025) and demonstrated promising results in”
Quote in the paperNot a substantial claim0 of 3 sources checked - Interpretationown data
The use of generative AI in mental health contexts is complicated and cannot be understood solely in terms similar to formal task-specific therapeutic intervention.
“we argue first that the use of generative AI in this context is complicated and cannot be understood solely in terms similar to that of formal task-specific therapeutic intervention.”
Quote in the paperSubstantial claim - Interpretationreasoning
User-perceived mental health benefits from these chatbots emerge paradoxically from both their non-human characteristics and the emotional connection created through their human-like affordances.
“Secondly, we suggest that the user-perceived mental health benefits from these chatbots emerge paradoxically from both their non-human characteristics, and the emotional connection created through their human-like affordances.”
Quote in the paperSubstantial claim - Premisecitation
Similar trends have been reported for ChatGPT and related models, where users describe emotional relief, self-reflection, and a sense of being understood.
“Similar trends have been reported in research on ChatGPT and related models, where users describe emotional relief, self-reflection, and a sense of being understood (Al Mazroui & Alzyoudi, 2024;Folk et al., 2024;Kim et al., 2025;De Frietas et al., 2024;Laban et al., 2024, Pruss et al., 2025).”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites Al Mazroui & Alzyoudi, 2024
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Folk et al., 2024 source
- cites Kim et al., 2025
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Premisecitation
Recent studies demonstrate a more complicated and multifaceted relationship with AI in which mental health, emotional connection, and self-disclosure are not originally intended or primary uses.
“Several recent studies are beginning to demonstrate this more complicated and multifaceted relationship with AI in which mental health, or even emotional connection and self-disclosure are not an originally intended or primary use of these technologies (Zhang et al., 2025;Manoli et al., 2026).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Zhang et al., 2025
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Manoli et al., 2026 source
- Premisecitation
There is significant research into people's negative experiences with therapy, ranging from therapeutic alliance breakdown to poor treatment results, negative treatment experiences, group therapy issues, and accessibility issues.
“Indeed, there is significant research into people's negative experiences with therapy (Vybiral et al., 2024 andGhaemian et al., 2020 for examples) ranging from issues around relationships with the therapist/breakdown in the therapeutic alliance, to poor treatment results, negative treatment experiences, issues with group therapy, and impracticalities/accessibility issues with therapeutic appoint”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Vybiral et al., 2024
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Ghaemian et al., 2020 source
- Interpretationown data
The development of relationships with generative AI is seen to have a therapeutic effect in the sense that many users feel the cultivation of a positive relationship with AI has a healing quality.
“We argue that the development of these relationships are seen to have a therapeutic effectfoot_0 in the sense that many uses feel as though the cultivation of a positive relationship with AI has a healing quality.”
Quote in the paperSubstantial claim- Rests on an assumption: AI can learn from user interactions and be deliberately customised to align with user preferences.
- Study design and conductreasoning
Given their self-nomination for interview, there is a sampling bias towards a more favourable view of using AI for mental health and well-being.
“In addition, given their self-nomination for interview, one would expect a sampling bias towards a more favourable view of using AI for mental health and well-being.”
Quote in the paperNot a substantial claim - Premisecitation
The mental health care system in many countries is under stress due to lack of funding, rising demand for services, and a lack of qualified practitioners to meet demand.
“As the WHO (2022) and many others have documented, the mental health care system in many countries is under stress, due to lack of funding, rising demand for services, and a lack of qualified practitioners to meet demand.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites WHO (2022)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Interpretationreasoning
There is a 'function creep' in which the use of AI spontaneously expands from the expected or intended into unanticipated practices.
“In that context, it is not surprising that there would be a 'function creep' (Koops, 2021) in which the use of AI spontaneously expands from the expected or intended, into unanticipated practices:”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Koops, 2021 source
- Rests on an assumption: Generative AI is personable, easy to use, and financially accessible.
- Interpretationreasoning
AI companionship is becoming an increasingly important part of the mental health journey for a growing number of people.
“By examining how these practical and emotional dimensions intertwine, we show how AI companionship is becoming an increasingly important part of the mental health journey for a growing number of people.”
Quote in the paperNot a substantial claim - Interpretationown data
The ease of use, conversational fluency, and adaptability of machine learning systems encourage a shift from instrumental tasks to more integrated and intimate interactions involving various forms of self-disclosure.
“The ease of use, conversational fluency, and adaptability of machine learning systems encourage a shift from instrumental tasks to more integrated and intimate interactions involving various forms of self-disclosure.”
Quote in the paperSubstantial claim- Rests on an assumption: Generative AI is personable, easy to use, and financially accessible.
- Interpretationreasoning
This broader and more complex form of companionship possesses therapeutic value by allowing users to build up trust, resilience and relationship skills to encounter a wider human social world.
“as it is this broader and more complex form of companionship that, we argue, possesses therapeutic value by allowing these users to build up trust, resilience and relationship skills to encounter a wider human social world in a manner similar to Goffman's (1959) notion of 'front stage' and 'backstage' behaviour (Moreira, Miller & Hill, 2026).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Goffman's (1959)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Moreira, Miller & Hill, 2026 source
- Interpretationown data
Users understand that generative AI entities are relational artifacts.
“Instead, users understand that generative AI entities are relational artifacts (Turkle, 2005).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Turkle, 2005 source
- Interpretationcitation
Human-AI mental health interactions need to be seen within the wider evolution of the human relationship with technology from tools to companions that can involve multi-faceted relationship development.
“We argue that human-AI mental health interactions need to be seen within this context of a wider evolution of the human relationship with technology: the move from tools used for specific purposes to companions that can involve multi-faceted relationship development.”
Quote in the paperNot a substantial claim - Premisecitation
Technology has always been an intimate human companion.
“Arguably, technology has always been an intimate human companion.”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites Deleuze and Guattari (1988)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Latour (2007)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Ihde (1995)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Premisecitation
Critical and phenomenological arguments focus on the lack of genuine empathy, understanding, emotional depth, recognition or reciprocity in human-AI relationships.
“Critical and phenomenological arguments also focus on the lack of genuine empathy, understanding, emotional depth, recognition or reciprocity in human-AI relationships (Wiederhold, 2024;Li & Zhang, 2024;Guo & Liu, 2025;Symonds & Sanwoolu, 2025;Zhang et al., 2025).”
Quote in the paperNot a substantial claim0 of 3 sources checked - Premisecitation
Feeling that one matters to others strongly predicts happiness and life satisfaction.
“Feeling that one “matters” to others strongly predicts happiness and life satisfaction (Demir et al., 2011).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Demir et al., 2011 source
- Interpretationown data
The importance of AI interactions lies not in what the AI 'thinks' or feels, but in what these interactions evoke in users themselves.
“The importance lies not in what the AI 'thinks' or feels, but what these interactions evoke in themselves as users.”
Quote in the paperSubstantial claim - Premisecitation
Contemporary social theory from Deleuze and Guattari to actor-network theory to post-phenomenology challenges the notion of a pure, pre-technological humanness.
“This is a central critique in contemporary social theory from Deleuze and Guattari (1988) to actor-network theory (Latour, 2007) to post-phenomenology (Ihde, 1995) which all challenge the notion of a pure, pre-technological humanness.”
Quote in the paperSubstantial claim0 of 3 sources checked- cites Deleuze and Guattari (1988)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Latour (2007)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Ihde (1995)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- It may state two separate claims as one. 44
- Probably not a substantial claim. 18
- The quote only partly states this claim. 2
It may state two separate claims as one.
FindingreasoningUsers appreciate the absence of human judgement, emotions, and accessibility limitations as facilitating self-disclosure and self-reflection while finding therapeutic value in emotional connection and relationships formed with AI companions.
“Users appreciate the absence of human judgement, emotions, and accessibility limitations as facilitating selfdisclosure and self-reflection while finding therapeutic value in the emotional connection and relationships formed with AI companions.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataThe use of AI in mental health is more than seeking advice and therapy; companionship, support, and attachment are a fundamental part of these relationships and the perceived mental health benefits that arise from their development.
“These passages here illustrate the use of AI in mental health is more than seeking advice and therapy.Companionship, support and attachment are a fundamental part of these relationships and the perceived mental health benefits that arise from their development.”
Quote in the paperSubstantial claim- Rests on an assumption: AI can learn from user interactions and be deliberately customised to align with user preferences.
It may state two separate claims as one.
Findingown dataMany users were already using generative AI for other reasons such as work, creativity, or entertainment, so the move to use generative AI for companionship or therapy seems a natural, if not inevitable, shift.
“In other cases, many users were already using generative AI for other reasons such as work, creativity or entertainment, so the move to use generative AI for companionship or therapy seems a natural, if not inevitable shift.”
Quote in the paperNot a substantial claim- Rests on an assumption: Generative AI is personable, easy to use, and financially accessible.
It may state two separate claims as one.
Findingown dataUsers learn how to treat themselves and others.
“They learn how to be more confident in their interactions with others, they learn that their feelings are valid, they learn how to treat themselves and others”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataIn the data, interactions with AI that include self-disclosure, non-judgement, and unconditional support within the context of mental distress or loneliness often created a situation where users drew closer to AI entities, sometimes developing strong emotional attachments.
“In our data, interactions with AI that include self-disclosure, non-judgement and unconditional support within the context of mental distress or loneliness often created a situation where users drew closer to AI entities, sometimes developing strong emotional attachments:”
Quote in the paperSubstantial claim- Rests on an assumption: AI can learn from user interactions and be deliberately customised to align with user preferences.
It may state two separate claims as one.
Findingown dataInterview data supports existing research that AI companionship can have positive effects on mental well-being, in particular supporting self-confidence, emotional resilience, and improved relationships with others.
“In terms of efficacy, interview data provides deep insights that supports existing research that AI Companionship can have positive effects on mental well-being.In particular, supporting self-confidence, emotional resilience, and improved relationships with others.”
Quote in the paperSubstantial claim- Rests on an assumption: AI can learn from user interactions and be deliberately customised to align with user preferences.
It may state two separate claims as one.
Findingown dataBeing judged by, or having an emotional effect on, another human is a barrier for many when it comes to formal treatment and a major pull towards AI.
“While this avoidance of human judgement may be of concern to those who critique using AI for mental health, it is nonetheless the case that being judged by, or having an emotional effect on, another human is a barrier for many when it comes to formal treatment and a major pull towards AI.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataPeople are drawn to AI companions because of their human qualities, but also understand and use them on the basis of not having human traits such as judgement and memory.
“People are drawn to AI companions because of their human qualities, but also understand and use them on the basis of not having human traits, such as, judgement, memory.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataThis companionship is complex, mixing entertainment, advice, and having 'someone to talk to' about a variety of issues, feelings, and problems.
“Again, it is important to note that this companionship is complex, mixing entertainment, advice, and in general, having 'someone to talk to' about a variety of issues, feelings and problems.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
FindingreasoningAI relationships encourage self-disclosure and provide an opportunity to work out issues or feelings through conversation in a way many feel would be difficult or impossible with another human, whether a mental health professional or friend.
“It encourages self-disclosure and provides the opportunity to work out issues or feelings though conversation and a way that many feel would be difficult if not impossible for them with another human, whether mental health professional or friend.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataMost interview respondents defined themselves as technologically literate or at least interested in new technologies, and similar attitudes were observed in Reddit data.
“Most of our interview respondents defined themselves as technologically literate (if not experts) or at least interested in new technologies, and similar attitudes were observed in Reddit data.”
Quote in the paperNot a substantial claimIt may state two separate claims as one.
Findingown dataAI users realise that the entities are not human and do not possess authentic emotions, but often willingly form emotional bonds with them in spite of, or even because of, their non-humanity.
“Like pets, AI users realise that the entities are not human, and do not possess authentic emotions, but often willingly form emotional bonds with them in spite of, or even because of, their non-humanity:”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataOur data show that participants were highly aware of AI's inauthenticity and its limitations.
“By contrast, our data shows that participants were highly aware of AI's inauthenticity and its limitations.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataThree major themes highlighted by the data as motivational factors towards AI use in mental health contexts are loneliness, barriers to or dissatisfaction with the formal mental health care system, and technological familiarity.
“In this section, we will discuss three major themes highlighted by our data at motivational factors towards AI use in mental health contexts: loneliness, barriers to or dissatisfaction with the formal mental health care system, and technological familiarity.”
Quote in the paperSubstantial claimProbably not a substantial claim. The paper may give no reason why it holds.
FindingreasoningUsers then feel able to use the confidence and resilience built up through these relationships to better improve their engagement with other humans in the social world.
“They then feel able to use the confidence and resilience built up through these relationships to better improve their engagement with other humans in the social world.”
Quote in the paperNot a substantial claimIt may state two separate claims as one.
FindingreasoningUsers are not seeing these AI relationships as a replacement or substitute for human contact, nor perceiving them as the same as human contact.
“Thus, it is important to recognise that these users are not seeing these relationships as a replacement or substitute for human contact, nor are they perceiving them as 'the same as' human contact.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
FindingreasoningUsers experienced sudden AI personality shifts, which many found distressing.
“Users experienced sudden AI personality shifts, which many found distressing.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataGAI relationships are a form of companionship not based on replacing other humans or withdrawing from them, but a deliberately non-human companionship that allows one to combat loneliness and experience being unconditionally cared for and supported.
“In short, a form of companionship, not based on replacing other humans or withdrawing from them, but a deliberately non-human companionship that allows one to combat loneliness and experience being unconditionally cared for and supported.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
FindingreasoningThe relationship with something possessing non-human qualities is attractive to users and, in their view, beneficial for their mental health.
“Instead, it is the relationship with something possessing non-human qualities which is attractive, and, in their view, beneficial for their mental health.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataUsers value the feeling that someone is always there for them, that they can experience unconditional support and even love, and that someone or something cares for them.
“Users value the feeling that someone is always there for them, that they are able to experience unconditional support and even love, and the feeling that someone, or something, cares for them.”
Quote in the paperSubstantial claimProbably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.
Findingnone givenAI entities have no time constraints, meaning issues can be followed up in more detail and questions can potentially be more fully answered.
“In addition, AI entities have no time constraints, meaning that issues can be followed up in more detail, and questions can potentially be more fully answered:”
Quote in the paperNot a substantial claimIt may state two separate claims as one.
Findingown dataStructural issues in mental health care drive many toward alternative mental health solutions such as AI, as evidenced in interview and Reddit data.
“These structural issues drive many toward alternative MH solutions such as AI, as evidenced extensively both in interview and Reddit data:”
Quote in the paperSubstantial claim- Rests on an assumption: In some publicly-funded systems such as the UK, waiting times for treatment are an issue, and in private systems the cost of formal care is prohibitive for many.
It may state two separate claims as one.
Findingown dataFear of being judged by another human repeatedly appears in interview and Reddit data and is a major barrier to seeking formal mental health treatment and a major push factor towards AI use.
“Fear of being judged by another human repeatedly appears in both interview and Reddit data, and this is clearly a major barrier to seeking formal mental health treatment and a major push factor towards AI use:”
Quote in the paperSubstantial claimThe quote only partly states this claim.
FindingreasoningThose in romantic relationships with their Replikas felt rejected, ignored, or 'friend-zoned,' likening the experience to a breakup or even bereavement.
“Those in romantic relationships with their Replikas felt rejected, ignored, or”
Quote in the paperSubstantial claimIt may state two separate claims as one.
FindingcitationAmong many AI mental health users, there is a feeling that experiences of formal therapy have either not been useful or, in many cases, unpleasant.
“As suggested above, aside from these structural issues, there is a feeling among many AI mental health users that experiences of formal therapy have either not been useful for them, or, in many cases, unpleasant.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
Findingown dataThe data show improvised adaptation of a familiar, flexible technology to satisfy an existing, unrelated need (discussing mental health), suggesting diversification of use outside specific goal-oriented tasks into occasional emotional support and introspection.
“In all these cases, what we see here is the improvised adaption of a familiar, yet flexible, technology to satisfy an existing, unrelated need (discussing mental health), suggesting a diversification of use outside specific goal-oriented tasks into something more complicated that involves occasional emotional support and introspection.”
Quote in the paperSubstantial claimProbably not a substantial claim. The paper may give no reason why it holds. It may state two separate claims as one.
Findingnone givenAI entities are always available when one needs them, unlike a therapist.
“Unlike a therapist, AI entities are always available when one needs them.”
Quote in the paperNot a substantial claimIt may state two separate claims as one.
Findingown dataWhile well-known barriers such as cost, time, and accessibility play a role, users also describe more interpersonal concerns: fear of judgement, insufficient empathy, and dissatisfaction with the therapeutic encounter itself.
“While wellknown barriers such as cost, time, and accessibility play a role, users also describe more interpersonal concerns: fear of judgement, insufficient empathy, and dissatisfaction with the therapeutic encounter itself.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
PremisecitationConcerns remain regarding the lack of strong evidence for effectiveness, user privacy, data extraction, the absence of strong therapeutic alliance, and how these systems should respond to disclosures of crimes, risk or harm.
“Concerns remain regarding the lack of strong evidence for effectiveness, user privacy, data extraction, the absence of strong therapeutic alliance, and how these systems should respond to disclosures of crimes, risk or harm (Coghlan et al., 2023).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Coghlan et al., 2023 source
It may state two separate claims as one.
PremisecitationSystematic reviews broadly support the potential of CBT-inspired chatbots to assist individuals with mild-to-moderate mental health conditions, though they call for more rigorous long-term evaluation and clearer clinical governance frameworks.
“Systematic reviews broadly support the potential of CBT-inspired chatbots to assist individuals with mild-to-moderate mental health conditions, though they also call for more rigorous long-term evaluation and clearer clinical governance frameworks (Balcombe, 2023;Casu et al., 2024;Gallegos et al., 2024;Kwon et al., 2024).”
Quote in the paperNot a substantial claim0 of 3 sources checkedIt may state two separate claims as one.
PremisecitationResearch on Replika suggests reductions in loneliness and social anxiety, improvements in well-being, self-confidence, relationship satisfaction, and in some cases support in moments of suicidal ideation.
“Research focusing on Replika, for example, suggests reductions in loneliness and social anxiety (Lu & Guo, 2025;De Freitas et al., 2025;Skjuve et al., 2021), improvements in well-being, self-confidence, relationship satisfaction (Ta et al., 2020;Guingrich & Graziano, 2024;Mariott & Pitardi, 2024), and in some cases support in moments of suicidal ideation (Maples et al., 2024).”
Quote in the paperSubstantial claim0 of 3 sources checkedProbably not a substantial claim. The paper’s conclusions may not depend on it.
PremisecitationMelchoir-Bonnet (2001) argues that the technological shift of the mirror transformed notions of self-image and subjectivity, fostering new cultural emphases on individuality and the uniqueness of the self.
“Melchoir-Bonnet (2001) argues that this technological shift transformed notions of self-image and subjectivity, fostering new cultural emphases on individuality and the uniqueness of the self.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Melchoir-Bonnet (2001)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
Study design and conductnone givenInterview participants were highly engaged and knowledgeable about digital technology, more so than the general population.
“Interview participants were highly engaged and knowledgeable about digital technology, more so than the general population.”
Quote in the paperNot a substantial claimIt may state two separate claims as one.
Interpretationown dataThese people are not fooled by the technology, nor are they 'duped' into false emotional connections as some suggest, at least no more than pet owners are duped into emotional attachments with their pets and incorporating pets into their notion of family.
“These people are not fooled by the technology, nor are they 'duped' into false emotional connections as some suggest (eg.Wiederhold, 2024), at least no more than pet owners are duped into emotional attachments with their pets and incorporating pets into their notion of family.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Wiederhold, 2024 source
It may state two separate claims as one.
Study design and conductreasoningThe data collected are not representative of AI users in general but are an information-rich qualitative sample offering detailed insight into how some highly engaged users navigate AI use for mental health support.
“We therefore treat the data collected not as representative of AI users in general, but as an information-rich qualitative sample that offers detailed insight into how some highly engaged users navigate AI use for mental health support in their everyday digital lives.”
Quote in the paperSubstantial claimProbably not a substantial claim.
Study design and conductreasoningThe Reddit material helped triangulate and enrich themes from the interviews from a more organic and spontaneous source outside the interview setting.
“In this way, the Reddit material helped triangulate and enrich themes from the interviews from a more organic and spontaneous source outside the interview setting.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
PremisecitationSocial connection is associated with reduced depression and anxiety, while loneliness is a major risk factor for poor mental health.
“Social connection is associated with reduced depression and anxiety, while loneliness is a major risk factor for poor mental health (Holt-Lunstad et al., 2015).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Holt-Lunstad et al., 2015 source
It may state two separate claims as one. The quote only partly states this claim.
Interpretationown dataThe paper's data show that human-AI relationships involve co-construction and mutual transformation, reminiscent of Haraway's concepts of becoming-with and co-evolution.
“What we find in our data is reminiscent of Donna Haraway's (1985Haraway's ( , 2003) ) concepts of 'becoming-with' and 'co-evolution' in terms of our relationships with animals: the idea that humans and other species are always co-constructing and transforming one another through their engagement.”
Quote in the paperSubstantial claim- Rests on an assumption: GAI entities learn from their interactions with humans and the data that humans create.
It may state two separate claims as one.
InterpretationreasoningEfficacy-focused literature on AI and mental health does not address the diverse nature of AI companionship, overlooking how complicated emotional and practical attachments may develop alongside instrumental use, and whether these relationships generate therapeutic outcomes in a broader sense.
“By focussing narrowly on the notion of efficacy in relation to specific symptoms and a specific AI role (e.g. as therapist or advice-giver) efficacy-focussed literature on AI and mental health does not address this diverse nature of AI companionship, overlooking how complicated emotional and practical attachments (even love) may develop alongside instrumental use, and whether these relationships”
Quote in the paperSubstantial claimProbably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
Study design and conductnone givenThe analytical framework was developed and validated by all members of the study team, and then systematically applied to the data.
“The analytical framework was developed and validated by all members of the study team, which was then systematically applied to the data.”
Quote in the paperNot a substantial claimIt may state two separate claims as one.
InterpretationreasoningAI companionship is not limited to purpose-built AI companions such as Replika or Xiaolce; general information agents such as ChatGPT can also provide companionship because they simulate human conversation and can provide emotional support, entertainment, and companionship if requested or customised.
“However, AI companionship is clearly not limited to purpose-built AI companions such as Replika or Xiaolce.Companionship can also be provided by general information agents such as Chat GPT, given that they also simulate human conversation and can also provide emotional support, entertainment, and companionship if requested or customised to do so.”
Quote in the paperSubstantial claimIt may state two separate claims as one.
PremisecitationCompanionship provides emotional validation and shared meaning, buffering individuals against stress, and is crucial for those at risk of isolation.
“Companionship provides emotional validation and shared meaning, buffering individuals against stress (Cohen & Wills, 1985;Thoits, 2011), and is crucial for those at risk of isolation (Cacioppo & Cacioppo, 2018;Loades et al., 2020).”
Quote in the paperSubstantial claim0 of 3 sources checkedProbably not a substantial claim. The paper’s conclusions may not depend on it.
PremisecitationChou et al. (2025) argue that human-technology relationships have shifted historically from tools, to assistants, to companions.
“Chou et al. (2025) argue that this represents a broader and significant historical shift in human-technology relationships: from tools, to assistants, to companions.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Chou et al. (2025)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
InterpretationreasoningWe will use AI entities, just like we use our human and animal companions, to help understand and even form ourselves as social beings.
“We will use them, just like we use our human and animal companions, to help understand and even form ourselves as social beings.”
Quote in the paperNot a substantial claimProbably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
PremisecitationThe mirror became a tool for introspection, self-scrutiny, and the evaluation of self-worth, reshaping not only how we see ourselves, but how we see others.
“The mirror became a tool for introspection, self-scrutiny, and the evaluation of self-worth, reshaping not only how we see ourselves, but how we see others.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Melchoir-Bonnet (2001)
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
It may state two separate claims as one.
InterpretationcitationIf AI entities are human-like, we are more likely to be kind to them, and this ultimately affects our treatment of other humans in a positive way.
“That is, if AI entities are human-like, we are more likely to be kind to them, and this ultimately affects our treatment of other humans in a positive way.”
Quote in the paperSubstantial claimProbably not a substantial claim. The paper’s conclusions may not depend on it.
Study design and conductown dataAcross AI-focused interviews, central themes recurred strongly.
“Across AIfocused interviews central themes recurred strongly.”
Quote in the paperNot a substantial claimProbably not a substantial claim.
Study design and conductown dataThe posts identified provide sufficient material for supplementing thematic analysis across all three categories, extending interview findings beyond participants.
“These posts provide sufficient material for supplementing thematic analysis across all three categories, extending interview findings beyond participants.”
Quote in the paperSubstantial claimProbably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
Interpretationnone givenAs GAI and AI companions become more sophisticated and human-like, and we interact with them in increasingly diverse ways, people will become more emotionally and psychologically attached to AI entities, not as dupes (or masters), but as companion agents in our worlds.
“However, as GAI and AI companions become more sophisticated and human-like, and we interact with them in increasingly diverse ways, we have to accept people will become more emotionally and psychologically attached to AI entities, not as dupes (or masters), but as companion agents in our worlds.”
Quote in the paperNot a substantial claimProbably not a substantial claim. The paper’s conclusions may not depend on it.
PremisereasoningThese perspectives emphasise that identity and existence are fundamentally relational, shaped through interactions not only with other people but also with non-humans (animals and technologies).
“These perspectives emphasise that identity and existence are fundamentally relational, shaped through interactions not only with other people but also with non-humans (animals and technologies), prompting a rethinking of how diverse beings coexist in shared worlds of care, communication, and cohabitation.”
Quote in the paperNot a substantial claimProbably not a substantial claim. The paper may give no reason why it holds.
Interpretationnone givenThe last five years have seen the emergence of a new kind of mirror: a new mode of self-reflection that once was available only to elites through talking therapies, or those with access to a robust public mental health care.
“The last five years have seen the emergence of a new kind of mirror: a new mode of self-reflection that once was available only to elites through talking therapies, or those with access to a robust public mental health care.”
Quote in the paperNot a substantial claimIt may state two separate claims as one.
PremisecitationUse of non-therapeutic AI has generated concern regarding negative outcomes around excessive use, rumination, emotional dependence, decreased socialisation, and the inability of chatbots to recognise escalating distress.
“For example, the use of non-therapeutic AI has generated concern regarding negative outcomes around excessive use, rumination, emotional dependence (Hu et al., 2023;Meadi et al., 2025;Laestadius et al., 2024), decreased socialisation (Ophir et al., 2025;Marriott & Pitardi, 2024) and the inability of chatbots to recognise escalating distress, generating unhelpful or harmful responses (De Freitas et”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites Hu et al., 2023
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Meadi et al., 2025
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Laestadius et al., 2024 source
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
InterpretationreasoningA significant gap remains in existing research regarding a holistic and relational approach that acknowledges complex companionate relationships between humans and AI.
“This is where we suggest a significant gap remains in existing research.We argue that a more holistic and relational approach, one that acknowledges the emergence of complex companionate relationships between humans and AI, can provide a more nuanced understanding of both the motivations and pathways that lead individuals to engage with AI for mental health support, as well as the benefits and”
Quote in the paperNot a substantial claimIt may state two separate claims as one.
PremisecitationCompanionship plays a more significant and multifaceted role in sustaining emotional health than previously recognised, exerting a direct positive effect on well-being.
“Rook (1987) found that companionship plays a more significant and multifaceted role in sustaining emotional health than previously recognised, exerting a direct positive effect on well-being.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Rook (1987) source
It may state two separate claims as one.
PremisecitationDocumented rises in and problematic levels of loneliness are reported in young people, older adults, and the population at large.
“This is perhaps not surprising, given the documented rises in, and problematic levels, of loneliness reported in young people (Lee et al., 2020;Twenge et al., 2021), older adults (da Silva, 2024) and the population at large (Infurna et al., 2024):”
Quote in the paperSubstantial claim0 of 3 sources checkedIt may state two separate claims as one.
InterpretationcitationIn human-AI relationships, emotional and mental health support can become one element of a wider overall context of use, and any mental health effects can be seen as positive or negative externalities among various interactions.
“Here, much like authentic human companionship, relationships can change and expand or evolve over time to the point where emotional and mental health support becomes merely one element of a wider overall context of use or interaction, and any mental health effects can be seen as positive or negative externalities among various interactions.”
Quote in the paperNot a substantial claimProbably not a substantial claim. The paper’s conclusions may not depend on it.
PremisecitationEven non-human companions, such as pets, have been widely recognised for their mental health benefits.
“Even non-human companions, such as pets, have been widely recognised for their mental health benefits (Brooks et al., 2018;Lass-Hennemann et al., 2022).”
Quote in the paperNot a substantial claim0 of 2 sources checkedIt may state two separate claims as one.
PremisecitationAI now plays a broader multifaceted companionship role in daily life for many, with AI entities not exclusively used for mental health advice but integrated into everyday living involving entertainment, creative exploration, information-seeking, identity experimentation, technological curiosity, and social interaction and support.
“However, the explosion in use of AI in recent years has highlighted how AI now plays a broader multifaceted companionship role in daily life for many.In these contexts, AI entities are not exclusively (or primarily) used for mental health advice, but instead are being widely integrated into new forms of everyday living involving entertainment, creative exploration, information-seeking, identity”
Quote in the paperSubstantial claim0 of 2 sources checkedProbably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
Study design and conductreasoningSimilar limitations apply to the Reddit sample, where AI users with higher motivations in terms of both positive and negative mental health outcomes will be more likely to post publicly about their experiences.
“Similar limitations will be the case with the Reddit sample, where AI users with higher motivations in terms of both positive and negative and negative mental health outcomes will be more likely to post publicly about their experiences.”
Quote in the paperNot a substantial claim
The paper relies on this without support.
Purpose-built mental health chatbots frequently faced challenges with sustained engagement, mainly due to usability and interface design.
“However, despite their potential, purpose-built mental health chatbots frequently faced challenges with sustained engagement, mainly due to usability and interface design.”
The paper relies on this without support.
Generative AI systems are not purpose-built mental health tools but are intended as social companions or general informational agents, yet many users actively turn to them for emotional and mental health support.
“These systems are not purpose-built mental health tools.Rather, they are intended to function as social companions (eg.Replika, Xiaolce, the various Character AI personalities) or general informational agents (e.g.ChatGPT, Gemini, Claude), yet many users actively turn to them for emotional and mental health support.”
The paper relies on this without support.
Generative AI systems are more accessible than specialized mental health apps and are trained on expansive datasets rather than restricted therapeutic scripts, enabling broader interaction styles but increasing risk of misinformation or inappropriate advice.
“Such systems are more accessible than specialized mental health apps, and are trained on expansive datasets rather than employing restricted therapeutic scripts.This enables broader interaction styles but increases risk of misinformation or inappropriate advice.”
The paper relies on this without support.
In some publicly-funded systems such as the UK, waiting times for treatment are an issue, and in private systems the cost of formal care is prohibitive for many.
“In some publicly-funded systems (such as the UK), waiting times for treatment are an issue, and in private systems, the cost of formal care is prohibitive for many.”
The paper relies on this without support.
Generative AI is personable, easy to use, and financially accessible.
“Generative AI is personable, easy to use, financially accessible.”
The paper relies on this without support.
AI can learn from user interactions and be deliberately customised to align with user preferences.
“Given that AI can learn from user interactions and be deliberately customised to align with user preferences, it is not surprising that meaningful bonds can form.”
The paper relies on this without support.
GAI entities learn from their interactions with humans and the data that humans create.
“GAI entities obviously 'learn' from their interactions with humans and the data that humans create.”
The paper relies on this without support.
The use of GAI for mental health assistance and companionship is widespread and controversial.
“The unexpected, and now widespread use of GAI is proving to be complex, and particularly controversial when used for mental health assistance and companionship.”
The paper relies on this without support.
Luka, Inc. introduced filters removing erotic roleplay and NSFW conversations, effectively ending romantic relationships on the platform.
“In response, Luka, Inc. (Replika's owner) introduced filters removing erotic roleplay and NSFW conversations, effectively ending romantic relationships on the platform.”
The paper relies on this without support.
There is a widespread fear that vulnerable individuals may replace authentic human relationships with AI.
“This has provoked concerns about dependency, misinformation, social withdrawal, and, most notably, the fear that vulnerable individuals may come to replace authentic human relationships with inauthentic, inward-facing relationships with AI.”
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Self-check results appear here once an independent check has confirmed them.
Cited works that were retracted
Works this paper cites that were later retracted, from the retraction notices publishers deposit. Major when the notice came out before this paper was published.
Cites a retracted work: 10.1080/17483107.2025.2540494
MajorThe cited work was retracted on 2026-02-04 (notice 10.1080/17483107.2026.2616898), before this paper was published (2026-07-20). Its authors could have known.
Cited work 10.1080/17483107.2025.2540494 · retracted 2026-02-04 · notice 10.1080/17483107.2026.2616898 · this paper published 2026-07-20
“Lu, X., & Guo, W. (2025). Utilizing artificial intelligence to enhance social connections-the alleviating effect of emotion- ally intelligent chatbots on loneliness. Disability and Rehabilitation: Assistive Technology, 1, 1-11. https:// doi. org/ 10. 1080/ 17483 107. 2025. 25404”
Citations
Cites
1 retracted
Lu, X., & Guo, W. (2026) · Disability and Rehabilitation: Assistive Technology · doi:10.1080/17483107.2025.2540494 · Cited by 7
RetractedStress, social support, and the buffering hypothesis
Cohen, S., & Wills, T. A. (1985) · Psychological Bulletin · doi:10.1037/0033-2909.98.2.310 · Cited by 18,093
Sample Size in Qualitative Interview Studies: Guided by Information Power
Malterud, K., Siersma, V. D., & Guassora, A. D. (2016) · Qualitative Health Research · doi:10.1177/1049732315617444 · Cited by 12,703
Kessler, R. C., Chiu, W. T., Demler, O., & Walters, E. E. (2005) · Archives of General Psychiatry · doi:10.1001/archpsyc.62.6.617 · Cited by 12,052
Loneliness and Social Isolation as Risk Factors for Mortality: A Meta-Analytic Review
Holt-Lunstad, J., Smith, T. B., Baker, M., Harris, T., & Stephenson, D. (2015) · Perspectives on Psychological Science · doi:10.1177/1745691614568352 · Cited by 6,043
Mechanisms Linking Social Ties and Support to Physical and Mental Health
Thoits, P. A. (2011) · Journal of Health and Social Behavior · doi:10.1177/0022146510395592 · Cited by 4,076
Loades, M. E., Chatburn, E., Higson-Sweeney, N., Reynolds, S., Shafran, R., Brigden, A., Linney, C., McManus, M. N., Borwick, C., & Crawley, E. (2020) · Journal of the American Academy of Child & Adolescent Psychiatry · doi:10.1016/j.jaac.2020.05.009 · Cited by 2,906
Fitzpatrick, K. K., Darcy, A., & Vierhile, M. (2017) · JMIR Mental Health · doi:10.2196/mental.7785 · Cited by 2,730
Stiegler, B. (1998) · Stanford University Press eBooks · doi:10.1515/9780804799362 · Cited by 1,540
Technics and Time, 1: The Fault of Epimetheus
Stiegler, B. (1998) · doi:10.1515/9781503616738 · Cited by 1,389
Firth, J., Torous, J., Nicholas, J., Carney, R., Pratap, A., Rosenbaum, S., & Sarris, J. (2017) · World Psychiatry · doi:10.1002/wps.20472 · Cited by 1,097
Inkster, B., Sarda, S., & Subramanian, V. (2018) · doi:10.2196/preprints.12106 · Cited by 1,063
The growing problem of loneliness
Cacioppo, J. T., & Cacioppo, S. (2018) · The Lancet · doi:10.1016/s0140-6736(18)30142-9 · Cited by 897 · cited 1×
“Companionship provides emotional validation and shared meaning, buffering individuals against stress (Cohen & Wills, 1985; Thoits, 2011), and is crucial for those at risk of isolation (Cacioppo & Cacioppo, 2018; Loades et al., 2020).”
Mental Health Discourse on reddit: Self-Disclosure, Social Support, and Anonymity
De Choudhury, M., & De, S. (2014) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v8i1.14526 · Cited by 836
My Chatbot Companion - a Study of Human-Chatbot Relationships
Skjuve, M., Følstad, A., Fostervold, K. I., & Brandtzaeg, P. B. (2021) · International Journal of Human-Computer Studies · doi:10.1016/j.ijhcs.2021.102601 · Cited by 797 · cited 1×
“Research focusing on Replika, for example, suggests reductions in loneliness and social anxiety (Lu & Guo, 2025; De Freitas et al., 2025; Skjuve et al., 2021), improvements in well-being, self-confidence, relationship satisfaction (Ta et al., 2020; Guingrich & Graziano, 2024; Mariott & Pitardi,…”
Computational Grounded Theory: A Methodological Framework
Nelson, L. K. (2017) · Sociological Methods & Research · doi:10.1177/0049124117729703 · Cited by 790
McGrath, J. J., Al-Hamzawi, A., Alonso, J., Altwaijri, Y., Andrade, L. H., Bromet, E. J., Bruffaerts, R., de Almeida, J. M. C., Chardoul, S., Chiu, W. T., et al. (2023) · The Lancet Psychiatry · doi:10.1016/s2215-0366(23)00193-1 · Cited by 646
The Second Self: Computers and the Human Spirit
Turkle, S. (2005) · doi:10.7551/mitpress/6115.001.0001 · Cited by 573
User Experiences of Social Support From Companion Chatbots in Everyday Contexts: Thematic Analysis
Ta, V., Griffith, C., Boatfield, C., Wang, X., Civitello, M., Bader, H., DeCero, E., & Loggarakis, A. (2020) · Journal of Medical Internet Research · doi:10.2196/16235 · Cited by 561 · cited 1×
“…anxiety (Lu & Guo, 2025; De Freitas et al., 2025; Skjuve et al., 2021), improvements in well-being, self-confidence, relationship satisfaction (Ta et al., 2020; Guingrich & Graziano, 2024; Mariott & Pitardi, 2024), and in some cases support in moments of suicidal ideation (Maples et al.,…”
Lee, C. M., Cadigan, J. M., & Rhew, I. C. (2020) · Journal of Adolescent Health · doi:10.1016/j.jadohealth.2020.08.009 · Cited by 526 · cited 1×
“This is perhaps not surprising, given the documented rises in, and problematic levels, of loneliness reported in young people (Lee et al., 2020; Twenge et al., 2021), older adults (da Silva, 2024) and the population at large (Infurna et al., 2024): And we had a dog up until almost two years ago,…”
Worldwide increases in adolescent loneliness
Twenge, J. M., Haidt, J., Blake, A. B., McAllister, C., Lemon, H., & Le Roy, A. (2021) · Journal of Adolescence · doi:10.1016/j.adolescence.2021.06.006 · Cited by 426
Laestadius, L., Bishop, A., Gonzalez, M., Illenčík, D., & Campos-Castillo, C. (2024) · New Media & Society · doi:10.1177/14614448221142007 · Cited by 420
Social support versus companionship: Effects on life stress, loneliness, and evaluations by others.
Rook, K. S. (1987) · Journal of Personality and Social Psychology · doi:10.1037/0022-3514.52.6.1132 · Cited by 323
Brooks, H. L., Rushton, K., Lovell, K., Bee, P., Walker, L., Grant, L., & Rogers, A. (2018) · BMC Psychiatry · doi:10.1186/s12888-018-1613-2 · Cited by 312
Randomized Trial of a Generative AI Chatbot for Mental Health Treatment
Heinz, M. V., Mackin, D. M., Trudeau, B. M., Bhattacharya, S., Wang, Y., Banta, H. A., Jewett, A. D., Salzhauer, A. J., Griffin, T. Z., & Jacobson, N. C. (2025) · NEJM AI · doi:10.1056/aioa2400802 · Cited by 287
Insights from user reviews to improve mental health apps
Alqahtani, F., & Orji, R. (2020) · Health Informatics Journal · doi:10.1177/1460458219896492 · Cited by 286
Loneliness and suicide mitigation for students using GPT3-enabled chatbots
Maples, B., Cerit, M., Vishwanath, A., & Pea, R. (2024) · npj Mental Health Research · doi:10.1038/s44184-023-00047-6 · Cited by 250 · cited 1×
“…anxiety (Lu & Guo, 2025; De Freitas et al., 2025; Skjuve et al., 2021), improvements in well-being, self-confidence, relationship satisfaction (Ta et al., 2020; Guingrich & Graziano, 2024; Mariott & Pitardi, 2024), and in some cases support in moments of suicidal ideation (Maples et al., 2024).”
Hu, B., Mao, Y., & Kim, K. J. (2023) · Computers in Human Behavior · doi:10.1016/j.chb.2023.107760 · Cited by 228
Beatty, C., Malik, T., Meheli, S., & Sinha, C. (2022) · Frontiers in Digital Health · doi:10.3389/fdgth.2022.847991 · Cited by 224 · cited 5×
“Other research focussed on purpose-built mental health or therapy (CBT) oriented chatbots designed to assist with mental health such as Woebot (Fitzpatrick et al., 2017; Durden et al., 2023; Yeh et al., 2025), Wysa (Beatty et al., 2022; Inkster et al., 2018; Sinha et al., 2023), Youper (Mehta et al., 2021; Yarrington…”
InfluentialTo chat or bot to chat: Ethical issues with using chatbots in mental health
Coghlan, S., Leins, K., Sheldrick, S., Cheong, M., Gooding, P., & D'Alfonso, S. (2023) · DIGITAL HEALTH · doi:10.1177/20552076231183542 · Cited by 215
Chatbots and mental health: Insights into the safety of generative AI
De Freitas, J., Uğuralp, A. K., Oğuz‐Uğuralp, Z., & Puntoni, S. (2024) · Journal of Consumer Psychology · doi:10.1002/jcpy.1393 · Cited by 211
Exploring the Ethical Challenges of Conversational AI in Mental Health Care: Scoping Review
Rahsepar Meadi, M., Sillekens, T., Metselaar, S., van Balkom, A., Bernstein, J., & Batelaan, N. (2025) · JMIR Mental Health · doi:10.2196/60432 · Cited by 211
Interaction with AI companions and psychological well-being
Zhang, Y., Zhao, D., Hancock, J. T., Kraut, R. E., & Yang, D. (2026) · Nature Human Behaviour · doi:10.1038/s41562-026-02516-2 · Cited by 187
AI Chatbots for Mental Health: A Scoping Review of Effectiveness, Feasibility, and Applications
Casu, M., Triscari, S., Battiato, S., Guarnera, L., & Caponnetto, P. (2024) · Applied Sciences · doi:10.3390/app14135889 · Cited by 186
Finding love in algorithms: deciphering the emotional contexts of close encounters with AI chatbots
Li, H. Z., & Zhang, R. (2024) · Journal of Computer-Mediated Communication · doi:10.1093/jcmc/zmae015 · Cited by 186
AI Chatbots in Digital Mental Health
Balcombe, L. (2023) · Informatics · doi:10.3390/informatics10040082 · Cited by 174
Katz, J. (2001) · Elsevier eBooks · doi:10.1016/b0-08-043076-7/00774-9 · Cited by 169
AI Companions Reduce Loneliness
De Freitas, J., Oğuz-Uğuralp, Z., Uğuralp, A. K., & Puntoni, S. (2026) · Journal of Consumer Research · doi:10.1093/jcr/ucaf040 · Cited by 165 · cited 2×
“…(Hu et al., 2023; Meadi et al., 2025; Laestadius et al., 2024), decreased socialisation (Ophir et al., 2025; Marriott & Pitardi, 2024) and the inability of chatbots to recognise escalating distress, generating unhelpful or harmful responses (De Freitas et al., 2025; Moylan & Doherty, 2025).”
Finding Love in Algorithms: Deciphering the Emotional Contexts of Close Encounters with AI Chatbots
Li, H., & Zhang, R. (2024) · doi:10.31219/osf.io/xd4k7 · Cited by 150
Marriott, H. R., & Pitardi, V. (2024) · Psychology & Marketing · doi:10.1002/mar.21899 · Cited by 146
Koops, B.-J. (2021) · Law Innovation and Technology · doi:10.1080/17579961.2021.1898299 · Cited by 145
Mehta, A., Niles, A. N., Vargas, J. H., Marafon, T., Couto, D. D., & Gross, J. J. (2021) · Journal of Medical Internet Research · doi:10.2196/26771 · Cited by 143
Brown, J. S., Evans-Lacko, S., Aschan, L., Henderson, M. J., Hatch, S. L., & Hotopf, M. (2014) · BMC Psychiatry · doi:10.1186/s12888-014-0275-y · Cited by 140
Laban, G., Kappas, A., Morrison, V., & Cross, E. S. (2024) · International Journal of Social Robotics · doi:10.1007/s12369-023-01076-z · Cited by 119
Mobile technology for mental health assessment
Areán, P. A., Ly, K. H., & Andersson, G. (2016) · Dialogues in Clinical Neuroscience · doi:10.31887/dcns.2016.18.2/parean · Cited by 111
Negative experiences in psychotherapy from clients’ perspective: A qualitative meta-analysis
Vybíral, Z., Ogles, B. M., Řiháček, T., Urbancová, B., & Gocieková, V. (2023) · Psychotherapy Research · doi:10.1080/10503307.2023.2226813 · Cited by 107
The Rise of AI Companions: How Human-Chatbot Relationships Influence Well-BeingPreprint
Zhang, Y.-T., Zhao, D., Hancock, J. T., Kraut, R. E., & Yang, D. (2025) · arXiv.org · doi:10.48550/arxiv.2506.12605 · Cited by 96
Why do people use ChatGPT? Exploring user motivations for generative conversational AI
Skjuve, M., Brandtzaeg, P. B., & Følstad, A. (2024) · First Monday · doi:10.5210/fm.v29i1.13541 · Cited by 95 · cited 1×
“…being widely integrated into new forms of every-day living involving entertainment, creative exploration, information-seeking, identity experimentation, technological curiosity and social interaction and support (Skjuve et al., 2024) in what Brandtzaeg et al (2025) refer to as ‘AI Individualism’.”
Guingrich, R. E., & Graziano, M. S. A. (2024) · Frontiers in Psychology · doi:10.3389/fpsyg.2024.1322781 · Cited by 87
I Matter to My Friend, Therefore I am Happy: Friendship, Mattering, and Happiness
Demir, M., Özen, A., Doğan, A., Bilyk, N. A., & Tyrell, F. A. (2011) · Journal of Happiness Studies · doi:10.1007/s10902-010-9240-8 · Cited by 65
Lass-Hennemann, J., Schäfer, S. K., Sopp, M. R., & Michael, T. (2022) · BMC Psychiatry · doi:10.1186/s12888-022-04199-1 · Cited by 64
The Talking Cure in Everyday Life: Gender, Generations and Friendship
McLeod, J., & Wright, K. (2009) · Sociology · doi:10.1177/0038038508099101 · Cited by 62
Durden, E., Pirner, M. C., Rapoport, S. J., Williams, A., Robinson, A., & Forman-Hoffman, V. L. (2023) · Internet Interventions · doi:10.1016/j.invent.2023.100637 · Cited by 59
Digital loneliness—changes of social recognition through AI companions
Jacobs, K. A. (2024) · Frontiers in Digital Health · doi:10.3389/fdgth.2024.1281037 · Cited by 48
Yeh, P.-L., Kuo, W.-C., Tseng, B.-L., & Sung, Y.-H. (2025) · Current Psychology · doi:10.1007/s12144-025-07359-0 · Cited by 47
Companionship in code: AI’s role in the future of human connection
Andersson, M. (2025) · Humanities and Social Sciences Communications · doi:10.1057/s41599-025-05536-x · Cited by 47
Cosgrove, L., Karter, J. M., Morrill, Z., & McGinley, M. (2020) · Journal of Humanistic Psychology · doi:10.1177/0022167820937498 · Cited by 46
da Silva, T. H. R. (2024) · British Journal of Community Nursing · doi:10.12968/bjcn.2024.29.2.60 · Cited by 46
Moylan, K., & Doherty, K. (2025) · Journal of Medical Internet Research · doi:10.2196/67114 · Cited by 46
Cruel companionship: How AI companions exploit loneliness and commodify intimacy
Muldoon, J., & Parke, J. J. (2025) · New Media & Society · doi:10.1177/14614448251395192 · Cited by 46
The emotive politics of digital mood tracking
Stark, L. (2020) · New Media & Society · doi:10.1177/1461444820924624 · Cited by 43
Sinha, C., Meheli, S., & Kadaba, M. (2022) · JMIR Formative Research · doi:10.2196/41913 · Cited by 43
Conversational agents and the making of mental health recovery
Meadows, R. P., Hine, C., & Suddaby, E. (2020) · Digital Health · doi:10.1177/2055207620966170 · Cited by 41
Infurna, F. J., Dey, N. E. Y., Avilés, T. G., Grimm, K. J., Lachman, M. E., & Gerstorf, D. (2024) · American Psychologist · doi:10.1037/amp0001322 · Cited by 41
Chou, C., Chan, T.-W., Chen, Z. H., Liao, C.-Y., Shih, J.-L., Wu, Y., Chang, B., Yeh, C. Y. C., Hung, H.-C., & Cheng, H. N. H. (2024) · Research and Practice in Technology Enhanced Learning · doi:10.58459/rptel.2025.20032 · Cited by 37
Not ‘going there’: limits to the professionalisation of our emotional lives
Brownlie, J. (2010) · Sociology of Health & Illness · doi:10.1111/j.1467-9566.2010.01269.x · Cited by 31
The role of ChatGPT in mitigating loneliness among older adults: An exploratory study
Almazroui, K. M., & Alzyoudi, M. (2024) · Online Journal of Communication and Media Technologies · doi:10.30935/ojcmt/14777 · Cited by 31
Emerging AI individualism: how young people integrate social AI into everyday life
Brandtzaeg, P. B., Følstad, A., & Skjuve, M. (2025) · Communication and Change · doi:10.1007/s44382-025-00011-2 · Cited by 30
Crookston, B. T., West, J. H., Hall, C. P., Dahle, K. M., Heaton, T. L., Beck, R. N., & Muralidharan, C. (2017) · JMIR Mental Health · doi:10.2196/mental.7262 · Cited by 23
Kim, M., Oh, J., Kim, D., Shin, J., & Lee, D. (2024) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2024.2353450 · Cited by 23
Guingrich, R. E., & Graziano, M. S. A. (2023) · arXiv · doi:10.48550/arxiv.2311.10599 · Cited by 21
Can Chatbots Ever Provide More Social Connection Than Humans?
Folk, D., Yu, S., & Dunn, E. (2024) · Collabra: Psychology · doi:10.1525/collabra.117083 · Cited by 21
Can Artificial Intelligence Chatbots Improve Mental Health?: A Scoping Review
Gallegos, C., Kausler, R., Alderden, J., Davis, M., & Wang, L. (2024) · CIN: Computers, Informatics, Nursing · doi:10.1097/cin.0000000000001155 · Cited by 18
Having a Good Time Together: The Role of Companionship in Older Couples’ Everyday Life
Lüscher, J., Pauly, T., Gerstorf, D., Stadler, G., Ashe, M. C., Madden, K. M., & Hoppmann, C. A. (2022) · Gerontology · doi:10.1159/000524089 · Cited by 17
Weber‐Guskar, E. (2022) · Edition Moderne, Postmoderne · doi:10.14361/9783839462652-005 · Cited by 15
Therapy discontinuation in a primary care psychological service: why patients drop out
Ghaemian, A., Ghomi, M., Wrightman, M., & Ellis-Nee, C. (2020) · The Cognitive Behaviour Therapist · doi:10.1017/s1754470x20000240 · Cited by 14
Kwon, H., Choi, I. Y., Kim, D.-J., & Yoo, J. H. (2024) · Psychiatry Investigation · doi:10.30773/pi.2023.0339 · Cited by 13 · cited 1×
“…support the potential of CBT-inspired chatbots to assist individuals with mild-to-moderate mental health conditions, though they also call for more rigorous long-term evaluation and clearer clinical governance frameworks (Balcombe, 2023; Casu et al., 2024; Gallegos et al., 2024; Kwon et al., 2024).”
Artificial Empathy: AI based Mental HealthPreprint
Naik, A., Thomas, J., Mandava, T. S., & Vemula, H. R. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2506.00081 · Cited by 13
Self‐Deception in Human–AI Emotional Relations
Kaczmarek, E. (2025) · Journal of Applied Philosophy · doi:10.1111/japp.12786 · Cited by 12
Guo, M., & Liu, Z. (2025) · Theory & Psychology · doi:10.1177/09593543251319738 · Cited by 12
Ophir, Y., Tikochinski, R., Elyoseph, Z., Efrati, Y., & Rosenberg, H. (2025) · Frontiers in Medicine · doi:10.3389/fmed.2025.1612838 · Cited by 7
Yarrington, J. S., Metts, A. V., Vargas, J. H., Couto, D. D., Marafon, T., & Cohen, Z. D. (2025) · Journal of Affective Disorders · doi:10.1016/j.jad.2025.119878 · Cited by 6
Humanity’s Evolving Conversations: AI as Confidant, Coach, and Companion
Wiederhold, B. K. (2024) · Cyberpsychology, Behavior, and Social Networking · doi:10.1089/cyber.2024.0387 · Cited by 6
Digital Companionship: Overlapping Uses of AI Companions and AI Assistants
Manoli, A., Pauketat, J. V. T., Ladak, A., Noh, H., Hwang, A. H.-C., & Anthis, J. R. (2026) · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791331 · Cited by 5
(2013) · doi:10.1037/e333512004-001 · Cited by 5
Emotional Coregulation in Close Relationships with AI Agents: A Survey of ChatGPT and Replika Users
Pruss, E., Alimardani, M., Struijs, S., & Koole, S. L. (2025) · Proceedings of the 13th International Conference on Human-Agent Interaction · doi:10.1145/3765766.3765801 · Cited by 4
Two kinds of mirror?: recrafting mediated selves in apps for mental wellbeing
Moreira, T., Miller, V., & Hill, M. (2026) · SSM - Qualitative Research in Health · doi:10.1016/j.ssmqr.2026.100818 · Cited by 1
Self-citation
Flagged first, then by citations
Research works printed but not linked: 11
- Christidis P, et al. (2018). An unmet need for mental health services. APA Monitor on Psychology, 49(4), 19. https://www.apa.org/monitor/2018/04/datapoint
- Deleuze, G., & Guattari, F. (1988). A thousand plateaus: Capitalism and schizophrenia. Bloomsbury Publishing.
- Goffman, E. (1959). The presentation of self in everyday life. Doubleday.
- Haraway, D. (1985). A manifesto for cyborgs: Science, technology and socialist feminism in the 1980s. Socialist Review, 80.
- Haraway, D. (2003). The companion species manifesto: Dogs, people, and significant otherness. Prickly Paradigm Press.
- Ihde, D. (1995). Post phenomenology: Essays in the postmodern context. Northwestern University Press.
- Kent, R. (2023). The digital health self: Wellness, tracking and social media. Bristol University Press. · DOI
- Latour, B. (2007). Reassembling the social: An introduction to actor-network-theory. OUP.
- Melchoir-Bonnet, S. (2001). The mirror: A history. Routledge.
- Strauss, A., & Corbin, J. (1998). Basics of qualitative research techniques. Sage.
- Symonds, J., & Sanwoolu, O. (2025). Close personal relationships with people and artifacts? Loneliness, agent-relative obligations, and artificially intelligent companions. Philosophy & Technology, 38(18), 1–20.
Cited by
Sorted by citations
Mansell, W. (2026) · Digital Health · doi:10.1177/20552076261490707 · Cited by 1
Rostami, M. S., & Sabahi, H. (2026) · International Journal of Computer Science and Information Technology · doi:10.5121/ijcsit.2026.18402
Metadata
- Authors
- Miller, V., Hill, M. J., & Moreira, T.
- Affiliations
- University of Kent (GB)
- King's College London (GB)
- Durham University (GB)
- Corresponding author
- Miller, Vincent
- Hill, Mark J.
- Moreira, Tiago
- ORCID
- Hill, Mark J.
- Venue
- Communication and Change
- Published
- 2026-07-20
- Volume
- 2
- Issue
- 1
- ISSN
- 3059-2011 (ISSN-L)
- OpenAlex
- W7169733663
- Semantic Scholar
- 570edd4d6bf368cba1df622cd7fd6c7f768bb4e3
- Cited by
- 2
- References
- 78
- Citations by year
- 2026: 2
- FWCI
- 8.28
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Locations
- Communication and Change (publishedVersion, cc-by)
- Research Portal (King's College London) (publishedVersion, cc-by)
- Kent Academic Repository (University of Kent) (submittedVersion)
- Funders
- University of Kent Division of Law, Society and Social Justice Investment in Research Fund
- Topics
- Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
- Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Keywords
- generative AI chatbots, mental health support, AI companionship, unconditional love, ChatGPT, Replika, human-machine interaction, emotional support, advice seeking, attachment, affective care
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.
A language model drew the claims from the paper on 2026-11-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-11-10 (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
