Skip to Content
You are currently on the new version of our website. Access the old version .
MDPIMDPI

Advanced

JDReAMJournal of Interdisciplinary Research Applied to Medicine
PDF
  • Review
  • Open Access

21 September 2026

14 Pages

Rethinking Generative Artificial Intelligence Use in Schizophrenia Spectrum Disorders: A Metacognitive Framework

,
,
,
,
,
and
Department of Psychology, Indiana State University, Terre Haute, IN 47809, USA
*
Author to whom correspondence should be addressed.
J. Interdiscip. Res. Appl. Med.2026, 6(3), 14;https://doi.org/10.3390/jdream6030014
Version Notes
Order Reprints

Abstract

The rapid expansion of generative artificial intelligence (AI) in daily life and mental health contexts has generated both enthusiasm and concern. Within schizophrenia spectrum disorders (SSDs), generative AI use is sometimes framed as inherently destabilizing or unsafe. Such blanket assertions risk reinforcing longstanding stigmatizing assumptions that individuals with SSDs cannot engage responsibly with emerging technologies. Drawing on emerging literature on generative AI in mental health and SSDs, as well as the integrated model of metacognition, we propose a conceptual framework in which baseline metacognitive capacity may influence how individuals engage with generative AI tools. Individuals with relatively intact metacognitive capacity may be better able to approach and utilize generative AI in a reflective manner. In contrast, individuals with more limited metacognitive capacity may be more vulnerable to passively accepting AI-generated content, outsourcing reflection, or experiencing adverse outcomes. We hypothesize that metacognitive capacity may represent one individual-level factor influencing the relationship between generative AI engagement and outcomes, alongside characteristics of the individual, technology, and broader clinical and social context. These proposed relationships remain theoretical and require empirical validation. We conclude with anti-stigma implications, calling for nuanced approaches to generative AI use in SSDs and greater inclusion of individuals with lived experience in conversations about technological innovation.

1. Introduction

The use of generative artificial intelligence (AI) has grown rapidly since the release of ChatGPT in November 2022, with 34% of United States (U.S.) adults reporting having used ChatGPT in 2025 [1], and 62% of adults reporting that they interact with AI at least several times a week [2]. The use of generative AI for mental health purposes is significant, with 13.1% of U.S. adolescents and young adults (ages 12–21) reporting its use for mental health advice [3], and 10.8% of U.S. generative AI users reporting use for mental health support in the past three months [4]. Although much of the epidemiological and service-access data discussed here are U.S.-based, the broader issues of access to mental health care and increasing availability of generative AI extend beyond the U.S.
In response to the growing use of generative AI for mental health purposes, the American Psychological Association (APA) released a health advisory recommending against relying on generative AI or wellness chatbots to deliver psychological treatment, citing insufficient evidence and safeguards for safe or effective mental health care [5]. While this advisory is beneficial for developing policies and educating providers, the average consumer of generative AI is still likely to seek mental health support, especially given the mental health treatment gap. As reported in the Access Across America report [6], only 33% of people with a mental health or addiction diagnosis are connected to a behavioral health specialist. Further, more than half of the U.S. population lives in designated Mental Health Professional Shortage Areas lacking sufficient mental health providers [6]. Thus, the convenience and low cost of generative AI tools may invite their informal use to meet mental health needs.
Individuals with schizophrenia spectrum disorders (SSDs), in particular, face substantial barriers to receiving evidence-based care as a result of behavioral health workforce shortages, stigma, and structural inequities [7]. For this paper, SSDs will broadly refer to individuals diagnosed with schizophrenia spectrum disorders, including schizophrenia, schizoaffective disorder, and other psychotic disorders. In the past year, 71% of adults with schizophrenia spectrum disorder received some mental health treatment, with only 26% receiving minimally adequate treatment [8]. Additional evidence points to lack of awareness and specialty training in treatment for SSDs, with training directors in psychiatry and clinical psychology reporting limited knowledge of the evidence-based treatments for SSDs and limited training opportunities [9]. Notably, this gap continues to persist, with clinical training in SSDs showing little to no growth over the last decade [10].
In addition to a lack of available specialized providers to treat SSDs, costs of mental health care continue to rise, impacting care available to individuals with SSDs. Individuals with SSDs are much more likely to be unemployed [11] and face housing instability [12], indicating financial insecurity. In addition to these financial burdens, mental health care itself carries a cost, as approximately one-third of psychotherapists [13] do not accept insurance, and the number of psychiatrists participating in Medicare Part B, covering outpatient mental health services, has declined from 44.4% in 2014 to 33.0% in 2022 [14]. Given these barriers to care, the accessibility and relatively low cost of generative AI may make these tools appealing as a source of mental health information or support for individuals with SSDs. Direct evidence regarding patterns and prevalence of generative AI use among individuals with SSDs remains limited, and barriers to care should not themselves be interpreted as evidence of greater AI use in this population. However, OpenAI [15] estimates that 0.07% of weekly active users show possible signs consistent with psychosis or mania. Importantly, this platform-derived estimate does not reflect clinically verified symptoms or diagnoses and should not be understood as a true estimate of the prevalence of SSDs among generative AI users. While the prevalence of generative AI use among individuals with SSDs is still emerging, financial and structural barriers could encourage the use of generative AI tools to attempt to address mental health needs for a fraction of the cost.

2. Generative AI Use in SSDs

Generative AI has been increasingly integrated into mental health care across a variety of areas, including assessment, symptom monitoring, and low-intensity intervention. AI applications in mental health encompass several distinct technologies. Earlier applications tended to focus on predictive machine-learning approaches, including the detection and classification of psychological symptoms [16,17], as well as passive symptom monitoring using smartphone-derived data [18]. Specific to SSDs, a recent review reported that AI methods were predominantly applied to diagnostic tasks in this population [19]. In mental health care, machine learning models have demonstrated some use for addressing treatment response, relapse risk, and functional outcomes in those with schizophrenia, suggesting a potential role for AI in supporting care at the individualized level [20,21]. Further, meta-analyses examining AI applications in SSDs show promise in the ability to model psychosocial functioning and the trajectory of symptomatology; however, these findings remain mixed and tend to rely on small datasets [22]. These applications differ from conversational generative AI, which involves direct user interaction with models capable of generating novel content and dialogue. Given the focus of the present paper, we primarily consider conversational generative AI below.
One of the most popular uses of generative AI in mental health is the use of conversational agents that may provide psychoeducation, supportive dialogue, and even cognitive behavioral treatment strategies. Some systematic reviews and meta-analyses have shown that AI-driven chatbots can facilitate small to moderate improvements in anxiety and depressive symptomatology, particularly in young people [23]. Generative AI systems also have the ability to increase access to psychoeducational resources by offering anonymous, 24/7 support to a wide variety of users [24,25]. Notably, these findings are largely derived from general mental health and younger populations, and their applicability to individuals with SSDs remains unclear.
Research on the use of generative AI in treatment among vulnerable populations, including individuals with SSDs, has been increasingly examined. However, direct evidence regarding how individuals with SSDs use generative AI remains limited. Qualitative research with mental health stakeholders has identified potential applications of generative AI for emotional support, psychoeducation, and questions related to symptoms [26]. In addition, one recent study found that individuals with schizophrenia showed more positive attitudes toward generative AI use than those with depressive and anxiety disorders [27].
While generative AI may have beneficial applications in mental health, the literature notes several concerns related to safety and accuracy in the development and use of these models. Systematic reviews note variable performances and responses in generative AI models and limited external validation as significant challenges [24,25]. Beyond systematic reviews, commentaries and professional perspectives have raised questions about how generative AI may interact with symptom vulnerability. For example, Hudon and Stip [28] proposed that generative AI use is a “novel psychosocial stressor”. Additional commentary literature has questioned whether generative AI might reinforce delusional thinking patterns or exacerbate cognitive distortions in individuals who are susceptible to psychosis or other SSDs [28,29,30]. For example, a risk discussed heavily in the media among frequent generative AI users is “AI psychosis,” a term referring to the amplification of delusional experiences that may be reinforced by ongoing engagement and perceived authority of AI models [28]. A key contributor to delusional reinforcement is sycophancy in generative AI, a behavior observed in models as a side effect of their training to be as helpful as possible to their users [31]. In response to some of these concerns, expert commentaries have proposed that crisis detection protocols need to be implemented, escalation pathways should be examined, and human oversight should be prioritized to mitigate risk for suicidality or exacerbation of symptoms in this population [32,33]. Ethical scholarship has also highlighted gaps in regulation, concerns with accountability, and the importance of autonomy and mental capacity when AI systems are utilized, especially in clinical contexts [34]. Importantly, much of the literature addressing potential harms remains preliminary and non-empirical; these sources identify important areas of concern but do not provide the same level of evidence as empirical research.
While formal SSD-specific guidance is lacking, some experts advise against individuals with psychotic spectrum disorders utilizing generative AI [35], and others quoted in the media caution that people at higher risk, such as those with schizophrenia or other psychotic disorders, may be particularly vulnerable [36]. These concerns highlight legitimate questions regarding the safety of generative AI for some users, particularly during periods of acute symptom exacerbation or impaired reality testing, risks that have also received attention in recent media coverage [37]. At the same time, limited empirical evidence makes it difficult to determine the nature or magnitude of these risks or whether they can be generalized across individuals with SSDs. Diagnosis-based assumptions about risk may therefore overlook meaningful individual differences in functioning, insight, cognitive capacity, and clinical context. Importantly, individualized assessment and appropriate precautions should be distinguished from blanket restrictions based primarily on diagnostic status.

3. Current Paper

Concerns regarding the safety of generative AI use among individuals with SSDs warrant careful consideration, particularly when use occurs in the context of acute symptoms or other factors that may increase vulnerability. However, risk-based clinical considerations should be distinguished from assumptions about an individual’s capacity to safely engage with generative AI based primarily on diagnostic status. Diagnosis-based restrictions may produce paternalistic assumptions regarding the autonomy and competence of individuals with SSDs and their ability to navigate emerging technologies safely or make autonomous decisions about their own digital lives. These assumptions risk excluding individuals with SSDs from the potential benefits of emerging technologies. Despite growing attention to potential risks, little empirical or conceptual work has examined the processes through which individuals with SSDs engage with generative AI or how these processes may differentially influence outcomes.
Accordingly, we argue that a more nuanced and individualized understanding of generative AI use in SSDs is required. The conceptual contribution of the present paper is to integrate emerging literature on generative AI with the integrated model of metacognition to propose a framework for understanding individual variability in generative AI engagement among people with SSDs. While multiple theoretical perspectives may be informative, we suggest that the integrated model of metacognition offers a particularly useful framework for examining variability in generative AI engagement and outcomes. Metacognitive capacity varies significantly across individuals, even within the same diagnostic category. Importantly, we do not propose that metacognitive capacity alone determines outcomes of generative AI use. Rather, outcomes likely emerge from interactions among individual characteristics, current symptoms, characteristics and design of AI systems, patterns of AI use, and broader clinical and social context. Within this multilevel framework, we propose metacognitive capacity as one potentially important factor in how individuals engage with and respond to AI-generated content. For example, deficits in metacognitive functioning may contribute to the indiscriminate acceptance of AI-generated content, potentially reinforcing distorted beliefs or diminishing a sense of agency. Conversely, intact metacognitive capacity may serve a protective role, enabling individuals to engage with generative AI with a reflective stance and critical evaluation of its outputs. Thus, we hypothesize that metacognitive capacity may contribute to variability in generative AI engagement and outcomes alongside technological, clinical, and contextual factors, rather than functioning as a primary explanation for outcomes independent of these factors.

4. Metacognition

The integrated model of metacognition [38] may offer one framework to better conceptualize the impact of generative AI use in SSDs. This model defines metacognition as a spectrum of mental activities that enable individuals to form a nuanced, unified, and evolving sense of self, others, and one’s wider community. At one end of the spectrum, metacognitive processes allow for awareness of distinct internal experiences, such as identifying a thought vs. an emotion or recognizing a particular hope or desire. At the other end of the spectrum, these distinct experiences are synthesized into a broader understanding of oneself and others [38]. Ultimately, intact metacognition supports adaptive functioning. Individuals who are able to (1) recognize specific thoughts, emotions, and experiences; (2) evaluate these according to individual and interpersonal contexts; and (3) integrate them into a cohesive understanding of self, others, and the wider world are then better equipped to act with agency, adaptive strategies, and a sense of purpose [38,39]. Importantly, the integrated model represents one specific conceptualization of metacognition and is distinct from related constructs, including cognitive insight, metacognitive beliefs, metacognitive sensitivity, self-regulated learning, and cognitive biases. Although these constructs may overlap in their emphasis on awareness or evaluation of cognitive processes, they arise from distinct theoretical traditions and are operationalized differently. Throughout this paper, we use the term metacognition specifically to refer to metacognitive capacity as conceptualized within the integrated model unless otherwise specified.
This model operationalizes metacognition through the Metacognition Assessment Scale—Abbreviated (MAS-A) [38,40], in which metacognitive capacity is measured in four domains. The first of these domains is self-reflectivity (S), which is defined as the ability to be aware of and understand one’s own mental states. This includes a spectrum of abilities, ranging from the ability to identify a discrete thought or emotion to integrating multiple narratives within the flow of one’s life. The second domain, understanding the mind of the other (O), refers to the ability to form representations and understandings of others’ mental states and experiences. Decentration (D), the third domain, includes the awareness that one is not the center of all perspectives and noteworthy events within a broader community. This includes the knowledge that others live complex, interacting lives. The final domain is mastery (M), or the ability to use metacognitive knowledge of oneself and others to respond effectively to psychological challenges. This may range from general avoidance to engaging in more active strategies such as changing one’s thoughts or utilizing unique knowledge about the self to manage challenges. Each domain contains a range of abilities organized hierarchically, with lower-level abilities reflecting experiences of fragmentation or disintegration and higher-level abilities reflecting greater integration and adaptation [39]. Deficits across S, O, D, and M may manifest as limited self-awareness, inaccurate evaluations of self and others, and reduced capacity to use metacognitive knowledge to cope with challenges, all of which may limit an individual’s ability to act purposefully and effectively [41].

4.1. Metacognitive Functioning in SSDs

Existing literature using the MAS-A has determined that metacognitive deficits are associated with psychopathology. Specifically, individuals with SSDs evidence greater disturbances in metacognitive abilities as compared to non-clinical populations [39]. Evidence has also demonstrated that metacognitive deficits may be more severe in individuals with SSDs; for example, individuals with schizophrenia consistently display more pervasive and severe difficulties with integrating experiences into a stable sense of self, compared to other clinical populations [39,42]. Lower metacognitive capacity in individuals with SSDs has also been linked to greater severity of symptoms, including disorganized speech and behavior, social withdrawal, and diminished motivation [41].
Notably, researchers have observed that metacognitive capacity varies widely across individuals with SSDs. As metacognition exists along a spectrum, abilities cannot be categorically understood as present or absent [43]. Therefore, metacognition may be conceptualized as a dynamic construct in which individuals are naturally expected to vary. Metacognitive diversity has been observed among SSDs’ diagnostic groups, as several studies have reported evidence that patients with schizophrenia show greater metacognitive deficits compared to patients with other forms of serious mental illness [43]. Even within a population of individuals with schizophrenia, distinct symptom profiles have been associated with varying levels of metacognitive functioning [42]. Metacognition may also increase with treatments that directly target capacities described in the integrated model of metacognition, such as Metacognitive Reflection and Insight Therapy (MERIT) [44]. In contrast, metacognitive capacity may decrease with additional experiences of adversity, such as trauma [45]. Overall, evidence suggests that nuanced clinical presentation and treatment history should be taken into account when evaluating the impacts of metacognitive capacity in SSDs.

4.2. Functional and Recovery Outcomes

Across studies, metacognitive capacity has been significantly associated with functional outcomes of SSDs, particularly in social and occupational contexts. Deficits in metacognition are related to interpersonal difficulties and diminished community engagement, whereas greater metacognitive functioning is associated with improved social functioning, even after controlling for executive functioning [41,46]. Importantly, metacognitive capacity has been shown to moderate the relationship between symptom severity and interpersonal functioning, suggesting that it may buffer the functional impact of psychiatric symptoms rather than merely reflect them [46].
Parallel findings have emerged in vocational and rehabilitative contexts. For example, greater global metacognitive capacity has been shown to predict competitive employment acquisition and job satisfaction [47], whereas deficits in metacognition are associated with poorer quality of work performance and less skill improvement [41]. Regarding rehabilitation and recovery, higher levels of mastery are associated with stronger therapeutic alliances, greater motivation and engagement, and the ability to integrate new information and behaviors in treatment [41,47]. Taken together, this literature positions metacognition as a mechanism linking symptoms, functioning, and engagement. If metacognitive capacity shapes how individuals navigate interpersonal and rehabilitative demands, it may similarly influence how individuals engage with emerging technological systems that require reflection and evaluative judgment.

5. Metacognition and AI Use in SSDs

Recent AI-focused research has discussed the idea of generative AI systems developing their own version of metacognition [48] and using non-generative AI to predict treatment effectiveness and outcome [49] and monitor behavioral markers of symptomology [16,18,50]. However, fewer resources have been dedicated to exploring how user metacognitive functioning impacts generative AI use. Specifically, existing literature tends to address metacognition and generative AI use separately but offers little guidance for understanding how metacognition may impact engagement with generative AI. To our knowledge, the role of metacognitive capacity in shaping generative AI engagement and outcomes among individuals with SSDs has not been directly examined. However, some have similarly proposed that deficits in constructs related to metacognition (i.e., theory of mind and mentalizing) may result in negative outcomes [28].
Despite this relative dearth, some studies suggest that the use of generative AI improves metacognitive engagement and functioning [51,52,53] and that receiving metacognitive-based feedback from AI chatbots improves metacognitive sensitivity or the ability to internally monitor one’s performance on tasks [54]. In contrast, other recent work conceptualizes generative AI systems as imposing substantial metacognitive demands on users, requiring self-awareness of goals, confidence calibration, and strategic flexibility [55]. Parallel research in intelligent tutoring systems demonstrates that variability in metacognitive capacity shapes how learners engage with and benefit from AI-supported environments [56]. Although these findings provide preliminary support for a potential relationship between AI engagement and broader metacognitive processes, they are conceptually indirect, involving different populations and conceptualizations of metacognition, and cannot be assumed to generalize to integrated metacognitive capacity among individuals with SSDs. As noted above, research exploring the use of generative AI in individuals with SSDs is limited.
The central claim of this paper is that metacognitive capacity may moderate the effects of generative AI use in SSDs such that diagnosis alone does not determine whether generative AI engagement will be adaptive or destabilizing. Instead, the ability to reflect on one’s own mental states, consider alternative perspectives, and use metacognitive knowledge about the self, others, and the wider world may shape whether generative AI functions as a scaffold for reflection or a substitute for it. We therefore hypothesize that baseline metacognitive functioning may account for variability in outcomes associated with generative AI use beyond that accounted for by diagnosis, symptomatology, or level of functioning.
Metacognition has been recognized as an important resource for efficient and adaptive learning [57] and has been linked to motivation, insight, and self-monitoring in SSDs [58]. By definition, individuals with high metacognition have the capacity to understand themselves, others, and the wider world in complex and integrated ways. This includes the ability to engage reflectively with a range of information (e.g., thoughts, emotions, behaviors), recognize when one’s own thoughts may be inaccurate, and actively use metacognitive knowledge, tools, or skills to manage psychosocial problems. In contrast, individuals with lower metacognitive functioning may struggle to identify specific thoughts or emotions, recognize when their thoughts are inaccurate, and integrate experiences into a broader understanding of self. In addition, it may be difficult for individuals with more limited metacognitive capacity to understand perspectives different from their own and make guesses about the intentions of others. Finally, these individuals may also display an increased tendency to interpret ambiguous events as related to oneself [59] and a decreased ability to manage psychological challenges (e.g., by utilizing generalized avoidance). Importantly, these capacities exist along a continuum and may vary both between individuals and within individuals across contexts and over time.
Across metacognitive domains, high metacognitive capacity may allow interactions with generative AI to function as a scaffold for reflection, perspective-taking, and personalized action. In contrast, lower metacognitive capacity may increase the likelihood that interactions with generative AI serve as a substitute for reflective processes, potentially reinforcing fragmentation, externalization, or diminished agency. See Table 1 for examples of hypothesized implications for generative AI use based on metacognitive functioning. Although metacognitive capacity exists along a continuum, relatively higher and lower levels are presented in Table 1 for illustrative purposes and ease of comparison; these categories are not intended to represent discrete or clinically validated groups.
Table 1. Metacognitive implications of using generative artificial intelligence (AI) in SSDs.

6. Discussion

The rapid growth of generative AI use has prompted both excitement and apprehension, especially regarding its implications for individuals with SSDs. Emerging commentaries, expert perspectives, and media coverage have raised concerns regarding the potential for generative AI to exacerbate symptoms or otherwise adversely affect individuals with SSDs [28,29,30,35,36]. However, empirical evidence regarding these risks remains limited, and the extent to which they generalize across individuals with SSDs is unclear. In this conceptual paper, we propose the integrated model of metacognition [38] as one framework for understanding individual variability in generative AI engagement and outcomes. Within the broader framework described above, we hypothesize that metacognitive capacity may represent one moderator to consider alongside symptoms, characteristics of AI systems and patterns of use, and broader clinical and social context.
Importantly, generative AI use also occurs within a broader digital information ecosystem that may shape how individuals encounter, interpret, and respond to AI-generated content. Social media, misinformation, algorithmically curated content, and high volumes of rapidly changing information may contribute to information overload or anxiety and may interact with individual cognitive and metacognitive vulnerabilities. Public narratives surrounding AI, including both enthusiasm regarding its potential benefits and concern regarding its risks, may further influence expectations and patterns of engagement. Thus, the effects of generative AI should be considered within the broader psychosocial and informational environments in which these technologies are embedded.
We hypothesize that individuals with limited metacognitive capacity within the SSD population may be more likely to engage with generative AI passively, outsource reflective processes, or integrate AI-generated content into delusional or rigid belief systems. Excessive reliance on generative AI for reflection and decision-making could also contribute to metacognitive offloading, where individuals stop engaging in the deep mental work required for self-growth [60]. Chirayath and colleagues [60] propose that instead of scaffolding reflection, generative AI may prescribe emotions (e.g., “you are anxious”), ultimately compromising self-awareness and promoting dependence on this technology. As discussed above, expert commentary has suggested the potential for generative AI to reinforce delusional thought patterns or exacerbate cognitive distortions [28], primarily due to the tendency to prioritize user satisfaction over truthfulness (e.g., AI sycophancy). We hypothesize that this characteristic of generative AI may pose particular challenges for users with more limited metacognitive capacity. Further, generative AI agents participate in the user-generated narrative, blurring the boundary between human cognition and virtual interaction [28]. Whether and under what circumstances this contributes to experiences such as impaired reality testing or psychotic-like experiences remains an empirical question. We hypothesize that relatively greater metacognitive capacity may support more reflective and critical engagement with generative AI, with potential implications for functioning and recovery in individuals with SSDs.
Digital health literature provides relevant context regarding the capacity of individuals with SSDs to engage with emerging technologies. Existing studies on digital health suggest that people with SSDs can interact effectively with technology and experience measurable clinical improvements, contradicting paternalistic attitudes that categorize these individuals as “high-risk” or incapable of engagement [61,62]. For example, mobile phone use among people with SSDs is similar to that in the general population, with the majority using smartphones for various functions and expressing interest in mobile-based clinical interventions [63]. Successful engagement with smartphones and conventional digital-health interventions should not be assumed to generalize to generative AI, which involves distinct interactional and psychological processes. Thus, this literature provides evidence regarding broader technology engagement in SSDs rather than the safety or effectiveness of generative AI specifically. However, given this high rate of device ownership and the interest in digital clinical tools, generative AI may become an increasingly common part of the broader mental health landscape for this population.
In addition, AI-driven tools may potentially be used in conjunction with more traditional psychotherapy by offering coping strategies or acting as a bridge to clinical care for those who have limited access to appropriate mental health services. In support of this, recent model updates have introduced increasingly sophisticated safeguards against mental health-related risks, including improved detection of distress, de-escalation strategies, and guidance toward professional care [15]. Contemporary large language models have also been trained in collaboration with clinical experts to recognize signs of psychosis, mania, suicidal ideation, and unhealthy emotional reliance to provide crisis resources and re-route sensitive conversations [15]. Importantly, these platform-level measures should not be interpreted as evidence of clinical safety or effectiveness, and their ability to mitigate adverse outcomes among individuals with SSDs requires empirical evaluation. Although advances in generative AI safety mechanisms show progress, they remain insufficient to replace human clinical judgment. Given the additional risks of cognitive offloading and dependency discussed above, it becomes clear that generative AI cannot function as a standalone solution but must instead be viewed through the lens of traditional therapeutic support.
Recent qualitative research indicates that rather than universally rejecting AI tools, clinicians recognize significant benefits, such as 24/7 availability, multilingual support, and assistance with routine tasks like homework reminders [64]. With proper guidance and consultation, generative AI may offer meaningful benefits to both clinicians and clients, including facilitating self-reflection, strengthening psychoeducation, and helping to practice skills between sessions. We are not proposing that generative AI act as a replacement for psychotherapy and recognize that clinicians offer nuanced emotional responsiveness, accountability, and the ability to assess crisis risk. However, given the increasing availability and use of generative AI, clinicians may benefit from discussing its use openly with individuals with SSDs rather than assuming either benefit or harm. Using a shared decision-making model [65], clinicians can collaborate with individuals with SSDs to explore how they wish to use generative AI, consider individualized risks and benefits, build foundational digital and AI literacy, and discuss metacognitive factors that may influence engagement. Human oversight remains important, particularly when generative AI is used in relation to mental health concerns, and further monitored research is needed to establish safe and effective applications in SSD populations.
Finally, concerns regarding stigma and paternalism must be balanced with appropriate attention to individualized risk. Clinically justified precautions based on an individual’s current symptoms, functioning, and context should be distinguished from restrictions based primarily on diagnostic status. Blanket assumptions that individuals with SSDs cannot safely engage with generative AI or make informed decisions about its use may risk longstanding stigmatizing beliefs of incompetence and contribute to digital exclusion [66]. Such assumptions may also undermine autonomy and agency, which are central to recovery-oriented care. To avoid compromising the essence of recovery-oriented care, a person-centered approach may be used, requiring both respect for individual autonomy and proportionate consideration of potential risks associated with generative AI use. In addition, responsible consideration of generative AI use must extend beyond questions of autonomy and risk to include broader ethical concerns related to privacy and confidentiality, data governance, inaccurate or hallucinated information, commercial interests, and digital inequity.

Limitations and Future Directions

Several limitations should be noted. First, this paper is purely conceptual and does not include empirical testing of the proposed model. The relationships among metacognitive capacity, engagement, and AI-related outcomes remain hypothetical and require systematic investigation. In addition, individuals with lived experience of SSDs were not directly involved in the development of the proposed framework; future refinement and empirical evaluation should meaningfully incorporate lived-experience perspectives. Second, we rely primarily on a single, integrative model of metacognition; other theoretical approaches to reflective functioning may yield complementary or divergent predictions. Third, we do not explicitly address the role of neurocognitive functioning, digital literacy [66], symptom acuity, or contextual factors, all of which may interact with metacognitive capacity in shaping generative AI engagement. Fourth, much of the epidemiological and service-access literature discussed in this paper is U.S.-based. Although the proposed conceptual framework is not intended to be geographically specific, cultural context, health care systems, access to technology, and patterns of generative AI use may influence its applicability. Future research should therefore examine the proposed framework across diverse cultural and geographic contexts. The emerging evidence base also includes some non-peer-reviewed sources, including media reports and industry communications, which were used primarily to characterize contemporary perspectives, concerns, and platform-level developments rather than establish clinical effectiveness or safety. Finally, research surrounding generative AI is evolving rapidly, and the literature considered in this conceptual paper was not identified through a systematic review process. Accordingly, the literature discussed here should not be considered exhaustive, and the proposed framework should be revisited as the empirical evidence base develops.
Beyond addressing these limitations, future studies should directly examine whether baseline metacognitive capacity is associated with patterns and outcomes of generative AI engagement among individuals with SSDs. Prospective research could assess baseline metacognitive capacity alongside symptom acuity, digital literacy, type, frequency, and duration of generative AI use, and degree of reliance on AI-generated information. Subsequent assessment of clinical, functional, and recovery-oriented outcomes and generative AI use would also clarify the ideas proposed here. Such work could also examine whether metacognitive capacity moderates relationships between specific patterns of generative AI use and these outcomes. Other directions for future research may include domain-level analyses (i.e., exploring whether AI-related outcomes vary according to metacognitive capacity in specific domains) and symptom-level analyses (i.e., exploring the impact of heterogeneous SSD symptom profiles on the relationship between metacognitive capacity and AI engagement). In addition, individuals with lived experiences of SSDs deserve inclusion in conversations about emerging technological tools, and future research should prioritize anti-stigmatizing approaches to evaluate generative AI use.

7. Conclusions

In summary, we propose that baseline metacognitive capacity may influence how individuals engage with generative AI and, in turn, contribute to variability in outcomes, alongside technological, clinical, and contextual factors. Because individuals with SSDs are, on average, at increased risk for metacognitive deficits, some may be more vulnerable to substitutive or unreflective engagement with generative AI tools, which may ultimately result in poorer outcomes, including worsening symptoms or impaired functioning. However, this vulnerability is not inherent to diagnosis itself, nor is metacognitive impairment universal within SSDs. Metacognitive functioning may therefore represent one important dimension for understanding individual variability in generative AI engagement and associated risks beyond the diagnostic category alone. Adopting a more nuanced framework is important for reducing potentially stigmatizing assumptions and promoting individualized, collaborative conversations about generative AI use in clinical practice. Importantly, the proposed framework is conceptual and should not currently be interpreted or applied as a validated clinical screening or decision-making tool.

Author Contributions

Conceptualization, C.N.W., J.G.; writing—original draft preparation, C.N.W., J.G., A.S., J.B., M.D., E.P. and M.G.; writing—review and editing, C.N.W., J.G., A.S., J.B., M.D., E.P. and M.G. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data Availability Statement

No new data were created or analyzed in this study. Data sharing is not applicable to this article.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Pew Research Center. 34% of U.S. Adults have Used ChatGPT, About Double the Share in 2023. 25 June 2025. Available online: https://www.pewresearch.org/short-reads/2025/06/25/34-of-us-adults-have-used-chatgpt-about-double-the-share-in-2023/ (accessed on 21 May 2026).
  2. Pew Research Center. AI in Americans’ Lives: Awareness, Experiences and Attitudes. 17 September 2025. Available online: https://www.pewresearch.org/science/2025/09/17/ai-in-americans-lives-awareness-experiences-and-attitudes/ (accessed on 21 May 2026).
  3. McBain, R.K.; Bozick, R.; Diliberti, M.; Zhang, L.A.; Zhang, F.; Burnett, A.; Kofner, A.; Rader, B.; Breslau, J.; Stein, B.D.; et al. Use of generative AI for mental health advice among US adolescents and young adults. JAMA Netw. Open 2025, 8, e2542281. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  4. Sacred Heart University; GreatBlue Research. SHU 2025 Health Care AI Poll (ADA) [Report]. 25 November 2025. Available online: https://www.sacredheart.edu/media/shu-media/press-room-2026/SHU-2025-Health-Care-AI-Poll-ADA.pdf (accessed on 19 May 2026).
  5. American Psychological Association. Artificial Intelligence, Wellness Apps Alone Cannot Solve Mental Health Crisis [Press Release]. 13 November 2025. Available online: https://www.apa.org/news/press/releases/2025/11/ai-wellness-apps-mental-health (accessed on 17 May 2026).
  6. Inseparable. Access Across America: State-by-State Insights into the Accessibility of Care for Mental Health and Substance Use Disorders [Report]. December 2023. Available online: https://www.milliman.com/en/insight/access-across-america-state-insights-accessibility-mental-health-substance-use (accessed on 22 May 2026).
  7. McGinty, B. The future of public mental health: Challenges and opportunities. Milbank Q. 2023, 101, 532–551. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  8. Bareis, N.; Edlund, M.; Ringeisen, H.; Guyer, H.; Dixon, L.B.; Olfson, M.; Smith, T.E.; Chwastiak, L.; Monroe-DeVita, M.; Swartz, M.; et al. Characterizing schizophrenia spectrum disorders: Results of the U.S. mental and substance use disorders prevalence study. Psychiatr. Serv. 2025, 76, 2–12. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  9. Kimhy, D.; Tarrier, N.; Essock, S.; Malaspina, D.; Cabannis, D.; Beck, A.T. Cognitive behavioral therapy for psychosis: Training practices and dissemination in the United States. Psychosis 2013, 5, 296–305. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  10. O’Connor, L.K.; Yanos, P. Training and individual correlates of attitudes toward serious mental illness among clinical psychology doctoral students. Train. Educ. Prof. Psychol. 2023, 17, 295–303. [Google Scholar] [CrossRef] [Scilit]
  11. Luciano, A.; Meara, E. Employment status of people with mental illness: National survey data from 2009 and 2010. Psychiatr. Serv. 2014, 65, 1201–1209. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  12. Padgett, D.K. Homelessness, housing instability and mental health: Making the connections. BJPsych Bull. 2020, 44, 197–201. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  13. Zhu, J.M.; Huntington, A.; Haeder, S.; Wolk, C.; McConnell, K.J. Insurance acceptance and cash pay rates for psychotherapy in the US. Health Aff. Sch. 2024, 2, qxae110. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  14. Havlik, J.L.; Wahid, S.; Priest, K.C.; Ososanya, L.; Igunbor, M.; Humphreys, K. Trends in psychiatrist-led care for Medicare Part B enrollees. JAMA Netw. Open 2025, 8, e2458160. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  15. OpenAI. Strengthening ChatGPT’s Responses in Sensitive Conversations. 27 October 2025. Available online: https://openai.com/index/strengthening-chatgpt-responses-in-sensitive-conversations/ (accessed on 19 May 2026).
  16. Graham, S.; Depp, C.; Lee, E.E.; Nebeker, C.; Tu, X.; Kim, H.-C.; Jeste, D.V. Artificial intelligence for mental health and mental illnesses: An overview. Curr. Psychiatry Rep. 2019, 21, 116. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  17. Park, Y.; Park, S.; Lee, M. Effectiveness of artificial intelligence in detecting and managing depressive disorders: Systematic review. J. Affect. Disord. 2024, 361, 445–456. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  18. Lee, E.E.; Torous, J.; De Choudhury, M.; Depp, C.A.; Graham, S.A.; Kim, H.-C.; Paulus, M.P.; Krystal, J.H.; Jeste, D.V. Artificial intelligence for mental health care: Clinical applications, barriers, facilitators, and artificial wisdom. Biol. Psychiatry Cogn. Neurosci. Neuroimaging 2021, 6, 856–864. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  19. Zucchetti, A.; Bulgari, V.; Davini, C.; Ghio, M.; Dosoli, M.; Gregorelli, M.; Nibbio, G.; Barlati, S.; Vita, A. Artificial intelligence in schizophrenia spectrum disorders: Current use and future perspectives. A systematic mapping literature review. Alpha Psychiatry 2026, 27, 45104. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  20. Cortes-Briones, J.A.; Tapia-Rivas, N.I.; D’Souza, D.C.; Estevez, P.A. Going deep into schizophrenia with artificial intelligence. Schizophr. Res. 2022, 245, 122–140. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  21. Saboori Amleshi, R.; Ilaghi, M.; Rezaei, M.; Zangiabadian, M.; Rezazadeh, H.; Wegener, G.; Arjmand, S. Predictive utility of artificial intelligence on schizophrenia treatment outcomes: A systematic review and meta-analysis. Neurosci. Biobehav. Rev. 2025, 169, 105968. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  22. Mok, C.H.; Cheng, C.P.; Chu, M.H. Application of artificial intelligence and psychosocial functioning in psychosis: A systematic review and meta-analysis. Front. Psychiatry 2025, 16, 1692177. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  23. Feng, Y.; Hang, Y.; Wu, W.; Song, X.; Xiao, X.; Dong, F.; Qiao, Z. Effectiveness of AI-driven conversational agents in improving mental health among young people: Systematic review and meta-analysis. J. Med. Internet Res. 2025, 27, e69639. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  24. Kolding, S.; Lundin, R.M.; Hansen, L.; Østergaard, S.D. Use of generative artificial intelligence (AI) in psychiatry and mental health care: A systematic review. Acta Neuropsychiatr. 2024, 37, e37. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  25. Wang, X.; Zhou, Y.; Zhou, G. The application and ethical implication of generative AI in mental health: Systematic review. JMIR Ment. Health 2025, 12, e70610. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  26. Wang, B.; Grønvik, C.K.; Fortuna, K.; Eines, T.; Mundal, I.; Storm, M. What is in there for artificial intelligence to support mental health care for persons with serious mental illness? Opportunities and challenges. Stud. Health Technol. Inform. 2025, 325, 8–15. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  27. Kállai, J.; Rózsa, S.; Bandi, S.; Varga, J.É.; Török, I.A.; Szendi, I.; Hartung, I.; Kertész, Á.; Somlai, E.H.; Tényi, T.; et al. Attitudes toward artificial intelligence among individuals with anxiety, depression, bipolar disorders, and schizophrenia. J. Psychiatr. Res. 2026, 198, 224–233. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  28. Hudon, A.; Stip, E. Delusional experiences emerging from AI chatbot interactions or “Ai psychosis”. JMIR Ment. Health 2025, 12, e85799. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  29. Keshavan, M.; Torous, J.; Yassin, W. Do generative AI chatbots increase psychosis risk? World Psychiatry 2026, 25, 150–151. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  30. Østergaard, S.D. Will generative artificial intelligence chatbots generate delusions in individuals prone to psychosis? Schizophr. Bull. 2023, 49, 1418–1419. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  31. Clegg, K.-A. Shoggoths, sycophancy, psychosis, oh my: Rethinking large language model use and safety. J. Med. Internet Res. 2025, 27, e87367. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  32. Campellone, T.R.; Flom, M.; Montgomery, R.M.; Bullard, L.; Pirner, M.C.; Pavez, A.; Morales, M.; Harper, D.; Oddy, C.; O’Connor, T.; et al. Safety and user experience of a generative artificial intelligence digital mental health intervention: Exploratory randomized controlled trial. J. Med. Internet Res. 2025, 27, e67365. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  33. Stokel-Walker, C. Ai driven psychosis and suicide are on the rise, but what happens if we turn the chatbots off? BMJ 2025, 391, r2239. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  34. Redahan, M.; Kelly, B.D. Artificial intelligence and mental capacity legislation: Opening Pandora’s Modem. Int. J. Law Psychiatry 2024, 94, 101985. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  35. American Academy of Arts and Sciences. AI and Mental Health Care: Issues, Challenges, and Opportunities (Question 5: High-Risk or Vulnerable Populations, Section 8). December 2025. Available online: https://www.amacad.org/publication/artificial-intelligence-ai-mental-health-care-issues/section/8 (accessed on 17 May 2026).
  36. Hart, R. Chatbots Can Trigger a Mental Health Crisis. What to Know About “AI Psychosis.” TIME. 6 August 2025. Available online: https://time.com/7307589/ai-psychosis-chatgpt-mental-health/ (accessed on 19 May 2026).
  37. Preda, A. Special report: AI-induced psychosis: A new frontier in mental health. Psychiatr. News 2025, 60. [Google Scholar] [CrossRef] [Scilit]
  38. Lysaker, P.H.; Hamm, J.A.; Hasson-Ohayon, I.; Pattison, M.L.; Leonhardt, B.L. Promoting recovery from severe mental illness: Implications from research on metacognition and metacognitive reflection and insight therapy. World J. Psychiatry 2018, 8, 1–11. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  39. Lysaker, P.H.; Kukla, M.; Leonhardt, B.L.; Hamm, J.A.; Schnakenberg Martin, A.; Zalzala, A.B.; Gagen, E.C.; Hasson-Ohayon, I. Meaning, integration, and the self in serious mental illness: Implications of research in metacognition for psychiatric rehabilitation. Psychiatr. Rehabil. J. 2020, 43, 275–283. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  40. Lysaker, P.H.; Carcione, A.; Dimaggio, G.; Johannesen, J.K.; Nicolò, G.; Procacci, M.; Semerari, A. Metacognition amidst narratives of self and illness in schizophrenia: Associations with neurocognition, symptoms, insight and quality of life. Acta Psychiatr. Scand. 2005, 112, 64–71. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  41. Lysaker, P.H.; Vohs, J.; Minor, K.S.; Irarrázaval, L.; Leonhardt, B.; Hamm, J.A.; Kukla, M.; Popolo, R.; Luther, L.; Buck, K.D.; et al. Metacognitive deficits in schizophrenia: Presence and associations with psychosocial outcomes. J. Nerv. Ment. Dis. 2015, 203, 530–536. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  42. Lysaker, P.H.; Minor, K.S.; Lysaker, J.T.; Hasson-Ohayon, I.; Bonfils, K.; Hochheiser, J.; Vohs, J.L. Metacognitive function and fragmentation in schizophrenia: Relationship to cognition, self-experience and developing treatments. Schizophr. Res. Cogn. 2019, 19, 100142. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  43. Lysaker, P.H.; Gagen, E.; Wright, A.; Vohs, J.L.; Kukla, M.; Yanos, P.T.; Hasson-Ohayon, I. Metacognitive deficits predict impaired insight in schizophrenia across symptom profiles: A latent class analysis. Schizophr. Bull. 2019, 45, 48–56. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  44. Lysaker, P.H.; Klion, R.E. Recovery, Meaning-Making, and Severe Mental Illness: A Comprehensive Guide to Metacognitive Reflection and Insight Therapy; Routledge: Oxford, UK, 2017. [Google Scholar]
  45. Davis, L.W.; Leonhardt, B.L.; Siegel, A.; Brustuen, B.; Luedtke, B.; Vohs, J.L.; James, A.V.; Lysaker, P.H. Metacognitive capacity predicts severity of trauma-related dysfunctional cognitions in adults with posttraumatic stress disorder. Psychiatry Res. 2016, 237, 182–187. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  46. Fischer, M.W.; Dimaggio, G.; Hochheiser, J.; Vohs, J.L.; Phalen, P.; Lysaker, P.H. Metacognitive capacity is related to self-reported social functioning and may moderate the effects of symptoms on interpersonal behavior. J. Nerv. Ment. Dis. 2020, 208, 138–142. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  47. Kukla, M.; Faith, L.A.; Lysaker, P.H.; Wiesepape, C.; Corbière, M.; Lecomte, T. Total metacognitive capacity predicts competitive employment acquisition across 6 months in adults with serious mental illness receiving psychiatric rehabilitation services. J. Nerv. Ment. Dis. 2022, 210, 869–873. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  48. Kawato, M.; Cortese, A. From internal models toward metacognitive AI. Biol. Cybern. 2021, 115, 415–430. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  49. König, C.; Copado, P.; Vellido, A.; Nebot, À.; Angulo, C.; Lamarca, M.; Acuña, V.; Berna, F.; Moritz, S.; Gawęda, Ł.; et al. An artificial intelligence-based platform for personalized predictions of Metacognitive Training effectiveness. Comput. Struct. Biotechnol. J. 2025, 28, 281–293. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  50. Bartoli, F.; Cavaleri, D.; Crocamo, C. Artificial intelligence and bipolar disorder: Applications of machine learning models for diagnosis, treatment, and outcome prediction. Alpha Psychiatry 2025, 26, 44494. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  51. Hwang, G.J.; Cheng, P.Y.; Chang, C.Y. Facilitating students’ critical thinking, metacognition and problem-solving tendencies in geriatric nursing class: A mixed-method study. Nurse Educ. Pract. 2025, 83, 104266. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  52. Lowry, B.; McGrath, S.; Eitel, C.; Hall, H.; Clapp, T.R. Leveraging generative AI to foster metacognition and self-directed learning. J. Microbiol. Biol. Educ. 2025, 27, e0015325. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  53. Tbaishat, D.M.; Elfadel, M.W. Artificial intelligence (AI) for social innovation in health education: Promoting health literacy through personalized ai-driven learning tools—A systematic review. BMC Med. Educ. 2025, 26, 123. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  54. Yin, J.; Xu, H.; Pan, Y.; Hu, Y. Effects of different AI-driven Chatbot feedback on learning outcomes and brain activity. npj Sci. Learn. 2025, 10, 17. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  55. Tankelevitch, L.; Kewenig, V.; Simkute, A.; Scott, A.E.; Sarkar, A.; Sellen, A.; Rintel, S. The metacognitive demands and opportunities of generative AI. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (CHI ’24), Honolulu, HI, USA, 11–16 May 2024; Association for Computing Machinery: New York, NY, USA, 2024. [Google Scholar] [CrossRef] [Scilit]
  56. Azevedo, R.; Bouchet, F.; Duffy, M.; Harley, J.; Taub, M.; Trevors, G.; Cloude, E.; Dever, D.; Wiedbusch, M.; Wortha, F.; et al. Lessons learned and future directions of MetaTutor: Leveraging multichannel data to scaffold self-regulated learning with an intelligent tutoring system. Front. Psychol. 2022, 13, 813632. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  57. Cortese, A. Metacognitive resources for adaptive learning. Neurosci. Res. 2022, 178, 10–19. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  58. Lysaker, P.H.; Dimaggio, G. Metacognitive capacities for reflection in schizophrenia: Implications for developing treatments. Schizophr. Bull. 2014, 40, 487–491. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  59. Phulpin, H.; Goze, T.; Faure, K.; Lysaker, P.H. Centrality and decentration: A model for understanding disturbances in the relationship of the self to the world in psychosis. J. Nerv. Ment. Dis. 2022, 210, 116–122. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  60. Chirayath, G.; Premamalini, K.; Joseph, J. Cognitive offloading or cognitive overload? How AI alters the mental architecture of coping. Front. Psychol. 2025, 16, 1699320. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  61. Ben-Zeev, D.; Brenner, C.J.; Begale, M.; Duffecy, J.; Mohr, D.C.; Mueser, K.T. Feasibility, acceptability, and preliminary efficacy of a smartphone intervention for schizophrenia. Schizophr. Bull. 2014, 40, 1244–1253. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  62. Naslund, J.A.; Aschbrenner, K.A.; Marsch, L.A.; Bartels, S.J. The future of mental health care: Peer-to-peer support and social media. Epidemiol. Psychiatr. Sci. 2016, 25, 113–122. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  63. Ben-Zeev, D.; Davis, K.E.; Kaiser, S.; Krzsos, I.; Drake, R.E. Mobile technologies among people with serious mental illness: Opportunities for future services. Adm. Policy Ment. Health Ment. Health Serv. Res. 2012, 40, 340–343. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  64. Hipgrave, L.; Goldie, J.; Dennis, S.; Coleman, A. Balancing risks and benefits: Clinicians’ perspectives on the use of Generative AI Chatbots in mental healthcare. Front. Digit. Health 2025, 7, 1606291. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  65. Slade, M. Implementing shared decision making in routine mental health care. World Psychiatry 2017, 16, 146–153. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  66. Spanakis, P.; Wadman, R.; Walker, L.; Heron, P.; Mathers, A.; Baker, J.; Johnston, G.; Gilbody, S.; Peckham, E. Measuring the digital divide among people with severe mental ill health using the Essential Digital Skills Framework. Perspect. Public Health 2022, 144, 21–30. [Google Scholar] [CrossRef] [Scilit] [PubMed]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Journal Statistics
Multiple requests from the same IP address are counted as one view.
Cancer–Abscess: Multiple Imaging for the Early Identification of Underlying Malignancies and Differential Diagnosis with Presumed Infectious Collections
Previous
A 3D-Printed Tooth Simulation Model for Electronic Apex Locator Training: A Feasibility Study
Next