Factors Predicting Intentions of Adoption and Continued Use of Artificial Intelligence Chatbots for Mental Health: Examining the Role of UTAUT Model,Stigma,Privacy Concerns,and Artificial Intelligence Hesitancy

Abstract

Background:

Artificial intelligence-based chatbots (AI chatbots) can potentially improve mental health care, yet factors predicting their adoption and continued use are unclear.

Methods:

We conducted an online survey with a sample of U.S. adults with symptoms of depression and anxiety (N = 393) in 2021 before the release of ChatGPT. We explored factors predicting the adoption and continued use of AI chatbots, including factors of the unified theory of acceptance and use of technology model, stigma, privacy concerns, and AI hesitancy.

Results:

Results from the regression indicated that for nonusers, performance expectancy, price value, descriptive norm, and psychological distress are positively related to the intention of adopting AI chatbots, while AI hesitancy and effort expectancy are negatively associated with adopting AI chatbots. For those with experience in using AI chatbots for mental health, performance expectancy, price value, descriptive norm, and injunctive norm are positively related to the intention of continuing to use AI chatbots.

Conclusions:

Understanding the adoption and continued use of AI chatbots among adults with symptoms of depression and anxiety is essential given that there is a widening gap in the supply and demand of care. AI chatbots provide new opportunities for quality care by supporting accessible, affordable, efficient, and personalized care. This study provides insights for developing and deploying AI chatbots such as ChatGPT in the context of mental health care. Findings could be used to design innovative interventions that encourage the adoption and continued use of AI chatbots among people with symptoms of depression and anxiety and who have difficulty accessing care.

Introduction

Anxiety and depression are among the most common mental health symptoms in the United States.^1,2 Despite their prevalence and negative impact, many individuals do not seek help due to stigma, lack of access, cost, and lack of awareness of available resources.^3

–8 Psychotherapy is usually viewed as an effective treatment for depression and anxiety. Treatments include cognitive behavioral therapy (CBT), positive psychology, problem-solving therapy, and acceptance and commitment therapy.^9
–11 While effective, the estimated treatment gap is 46.2% for anxiety disorders and 57.9% for affective disorders in North America.¹²

One of the newest developments in the digital mental health field is the emergence of artificial intelligence-based chatbots (AI chatbots),^13
–15 which are gaining recognition with the release of ChatGPT.^16
–18 AI chatbots are conversational or relational large language model programs that imitate human conversations as they naturally occur, and they have been broadly implemented in e-commerce, health care, and entertainment.¹³

Particularly, beginning with a computer program, known as ELIZA, developed to emulate the conversational abilities of a psychotherapist, AI chatbots have become increasingly prevalent in modern psychiatry.¹⁴ Commonly integrated with CBT,¹⁹ AI chatbots vary in complexity and are embedded into diverse forms, including mobile apps, websites, and SMS texting.²⁰

The most prominent example of AI chatbots for mental health is Woebot, which delivers CBT through instant messaging and mimics the relationship between human clinicians and patients.²¹ A review found that AI chatbots effectively improved depression, distress, stress, and acrophobia.²²

In some studies, AI chatbots have been found to be as effective as, or even more effective than, traditional mental health interventions as patients are more likely to self-disclose with AI chatbots.²³ Moreover, AI chatbots offer advantages such as accessibility, affordability, anonymity, and convenience.^24,25

Despite AI chatbots' potential to bridge the gap between supply and demand for care, there are few theoretically driven studies regarding the acceptability and continued use of AI chatbots among individuals with symptoms of depression and anxiety. Notably, fewer studies have examined the factors related to the continued use of AI chatbots.^26
–28

As one major issue plaguing health technology use is the abandonment of technology after initial adoption,^29
–31 it is worth examining factors that predict the intention to continue using AI chatbots to further our understanding of AI chatbot usage. This study aimed to identify a range of psychological and technological factors influencing the intention to adopt as well as to continue using AI chatbots, informed by literature on technology adoption, AI perception, and mental health help-seeking behavior.

Previous studies focused on the initial adoption of AI chatbots,^15,21,32 commonly framed by general technology acceptance models (e.g., technology acceptance model, AI device use acceptance model, and unified theory of acceptance and use of technology).^32

–35 Specifically, researchers found that performance expectancy, effort expectancy, and social influence such as descriptive and injunctive norms^36
–38 positively related to the intention to use coaching, medical, or mental health chatbots^32,33,37 among general adults,³⁷ LGBTQIA+ individuals,³³ and cancer patients.³²

Additionally, while there is less direct evidence, previous studies in the health context have repeatedly found that facilitating conditions^39
–41 and price values^26,42,43 are positively related to behavioral intention to use various health-related technologies. Considering that these factors are expected to positively predict the adoption and continued use of AI chatbots for mental health, we propose the hypotheses below:

H1: (a) Performance expectancy, (b) effort expectancy, (c) facilitating conditions, (d) price value, (e) descriptive norm, and (f) injunctive norm positively predict the intention of adoption and continued use of AI chatbots for mental health.

For incorporating new technologies such as AI into health care, people may exhibit ambivalence about using AI-driven chatbots, given the concerns around the quality, trustworthiness, and accuracy of health information provided by chatbots and the perceived lack of empathy from depersonalized chatbots, which researchers have termed as AI hesitancy to describe the phenomenon.⁴⁴

Relatedly, privacy concerns about access to personal data have long been established as a barrier to seeking help online for mental health problems, as commonly discussed in studies on digital mental health.^45
–47 Therefore, to further identify factors particularly relevant to AI technology, we propose the following hypotheses.

H2: (a) AI hesitancy and (b) privacy concerns will negatively predict the intention of adoption and continued use of AI chatbots for mental health.

Specific to the context of mental health help-seeking behavior, one major barrier is stigma, which can usually be categorized into self-stigma and social stigma.⁴⁸ Self-stigma refers to when an individual internalizes the negative attitudes people have toward mental health problems, whereas social stigma refers to the perceived public stigma associated with seeking professional help.⁴⁹

Researchers reported that stigma about mental illness or mental health services is significantly related to a reluctance of seeking help for mental health problems in the general population.⁴⁹ In contrast, a higher level of psychological distress is positively associated with general help-seeking behavior for emotional problems.⁵⁰ Therefore, we hypothesize that these factors also apply to using AI chatbots for mental health.

H3: (a) Self-stigma and (b) social stigma negatively predict the intention of adoption and continued use of AI chatbots for mental health, while (c) psychological distress positively predicts the intention of adoption and continued use of AI chatbots for mental health.

Moreover, AI hesitancy may interact with social stigma and privacy concerns, exacerbating or alleviating their effect on adoption and continued use. On the one hand, as AI systems require a vast amount of personal data, users may be concerned with the potential leakage of their personal information.⁵¹ In this case, AI hesitancy may exacerbate the effect of privacy concerns.

On the other hand, the lack of human-to-human interaction and the relatively high level of anonymity that AI chatbots afford may reduce privacy concerns.^24,25 Therefore, lower AI hesitancy may alleviate the effect of privacy concerns and encourage the adoption or continued use of mental health chatbots. Similarly, AI hesitancy may also impact social stigma's effect on using AI chatbots.

For those with lower AI hesitancy, the anonymity afforded by AI chatbots may lower the barrier of social stigma and therefore motivate the use of AI chatbots. For example, one study found that chatbot acceptability was significantly higher for stigmatized health issues.⁵² Therefore, we ask the following research question:

RQ1: Will AI hesitancy moderate the relationship between (a) privacy concern and (b) social stigma and the intention of adoption and continued use of AI chatbots for mental health?

Methods

PARTICIPANTS AND PROCEDURES

The institutional review board approved the study at the university where the study was conducted. We conducted an online survey among residents of the United Sates on Amazon Mechanical Turk (Mturk). Data were collected online using Qualtrics in June 2021. To screen individuals with symptoms of depression and anxiety, participants first completed the Patient Health Questionnaire 2 (PHQ-2) and then the Generalized Anxiety Disorder Questionnaire (GAD-2).

For all four statements, participants indicated their frequency of distress (not at all = 0, several days = 1, more than half the days = 2, and nearly every day = 3). The survey was discontinued for individuals who scored less than 3 points as they did not meet our criteria.

Individuals who passed the screening were presented with a consent form for the survey. After providing consent, they were presented with a broad definition of automated e-mental health tools, encompassing AI chatbots: “Automated e-mental health tools for anxiety and depression are automated mental health services and information delivered through the Internet and related technologies with professional input from therapists/clinicians, but without real-time human support. It does not include online therapy in which you talk to a human provider either in real time or asynchronously.”

After reading the definition, they were shown an embedded video introducing Woebot as an example of AI chatbots, followed by a question to indicate whether they have used similar AI chatbots before. Participants who answered “yes” were directed to survey questions that were phrased accordingly to examine continued use intention, and participants who answered “no” were directed to survey questions that were phrased to examine adoption intention.

After deleting responses that failed to pass half of the six attention check questions, our final sample consisted of 393 participants, with 208 participants who had not used AI chatbots before (AI nonusers) and 185 participants who had used AI chatbots before (AI users). The survey took about 25 minutes, and participants who completed the survey were paid $3 through Mturk.

MEASURES

To measure intention to use/continue using AI chatbots, we adopted the validated scale of behavioral intention based on measures of the UTAUT model.⁵³ Participants rated their intention to use/continue using AI chatbots by indicating their level of agreement on three statements on 7-point Likert scales (1 = strongly disagree and 7 = strongly agree). Cronbach's α was 0.96 for intention to use among nonusers and 0.79 for AI users. Scores were averaged across the three statements to create the score for intention to use/continue using AI chatbots.

Performance expectancy, effort expectancy, facilitating conditions, and price value were measured by adapting the corresponding scales of performance expectancy (⍶ = 0.72 for AI users and ⍶ = 0.90 for AI nonusers), effort expectancy (⍶ = 0.71 for AI users and ⍶ = 0.88 for AI nonusers), facilitating conditions (⍶ = 0.71 for AI users and ⍶ = 0.77 for AI nonusers), and price value (⍶ = 0.71 for AI users and ⍶ = 0.87 for AI nonusers) based on measures of the UTAUT model.

The statements were phrased in the present tense for AI chatbot users and in future tense for AI nonusers. Descriptive and injunctive norms were both measured by a two-item scale adapted from the scale of descriptive/injunctive norms based on previous research.⁵⁴

Self-stigma was measured using a validated 10-item scale of self-stigma of seeking help (⍶ = 0.85).⁵⁵ Social stigma was measured by a validated 5-item Social Stigma Scale for Receiving Psychological Help.⁵⁶ Privacy concern was measured by a three-item scale adapted from previous scales related to online information privacy (⍶ = 0.83).^57,58 AI hesitancy was measured using four items (⍶ = 0.80) adjusted from a technophobia scale,⁵⁹ such as “I prefer to have people handle my mental health issues than automated e-health tools.”

Psychological distress was measured by combining participants' scores on PHQ-2 and GAD-2.^60,61 We also measured demographic variables, including age, sex, gender, income, education, and race, for sample description. Participants' demographics are reported in Table 1.

Table 1.

Participant Demographics

N = 393	n (%)	M	SD
Age, years		38.46	1.10
Gender
Male	143 (36.4)
Female	250 (63.6)
Race
White	255 (64.9)
Black or African American	74 (18.8)
American Indian or Alaska Native	6 (1.5)
Asian or Pacific Islander	47 (12.0)
Mixed race	9 (2.3)
Other	2 (.5)
Education
High school degree	22 (5.6)
Some college or associate degree	41 (1.4)
Bachelor's degree	229 (58.3)
Beyond bachelor's degree	101 (25.7)
Income
Less than $39,999	116 (29.5)
$40,000 to $69,999	156 (39.7)
$70,000 to $99,999	89 (22.6)
$100,000 or more	32 (8.1)
Marital status
Married or living as married	304 (77.4)
Not married	89 (22.6)

Two hundred eight participants have not used AI tools before and 185 have used AI tools in the past. N = 192 for age. N = 193 for all other variables. Not married included those who are single, divorced, separated, or widowed.

AI, artificial intelligence; M, mean; SD, standard deviation.

Results

Two regression models with the intention to use or continue using AI chatbots as the dependent variables were tested. No multicollinearity issue was found, and all analyses were conducted in Stata 14. We reported the descriptive statistics and correlations of all relevant variables in Table 2 for AI nonusers and Table 3 for AI users. Regression model results are presented in Table 4.

Table 2.

Means, Standard Deviations, and Zero-Order Correlations of Relevant Variables for Artificial Intelligence Nonusers (N = 208)

	1	2	3	4	5	6	7	8	9	10	11	12
Intention to use	1
Performance expectancy	0.73^***	1
Effort expectancy	0.26^***	0.43^***	1
Facilitating conditions	0.30^***	0.42^***	0.73^***	1
Price value	0.65^***	0.69^***	0.44^***	0.45^***	1
Injunctive norms	0.46^***	0.58^***	0.47^***	0.59^***	0.60^***	1
Descriptive norms	0.65^***	0.57^***	0.28^***	0.22^**	0.62^***	0.45^***	1
Self-stigma	0.09	−0.01	−0.23^***	−0.28^***	0.02	−0.20^**	0.23^**	1
Social stigma	0.20^**	0.04	−0.09	−0.09	0.05	−0.08	0.27^***	0.64^***	1
AI hesitancy	0.08	−0.006	−0.05	−0.10	0.08	−0.04	0.42^***	0.36^***	0.42^***	1
Privacy concerns	0.03	−0.05	−0.09	−0.04	−0.08	−0.16^*	0.10	0.14^*	0.39^***	0.37^***	1
Psychological distress	0.21^**	0.04	0.01	0.02	0.03	0.05	0.23^***	0.28^***	0.33^***	0.23^***	0.05	1
M	4.63	5.08	5.48	5.46	5.09	5.25	4.06	3.34	4.32	4.56	4.64	6.16
SD	1.68	1.29	0.94	0.95	1.16	1.22	1.85	1.13	1.48	1.26	1.43	2.41

p < 0.05; ^** p < 0.01; ^*** p < 0.001.

Table 3.

Means, Standard Deviations, and Zero-Order Correlations of Relevant Variables for Artificial Intelligence Users (N = 185)

	1	2	3	4	5	6	7	8	9	10	11	12
Intention to continue use	1
Performance expectancy	0.85^***	1
Effort expectancy	0.77^***	0.80^***	1
Facilitating conditions	0.72^***	0.76^***	0.86^***	1
Price value	0.83^***	0.82^***	0.84^***	0.82^***	1
Injunctive norms	0.78^***	0.74^***	0.81^***	0.81^***	0.84^***	1
Descriptive norms	0.77^***	0.70^***	0.69^***	0.69^***	0.78^***	0.77^***	1
Self-stigma	0.07	0.001	−0.07	−0.02	0.01	0.06	0.23^**	1
Social stigma	0.69^***	0.67^***	0.69^***	0.77^***	0.75^***	0.74^***	0.76^***	0.35^***	1
AI hesitancy	0.63^***	0.63^***	0.66^***	0.75^***	0.72^***	0.73^***	0.72^***	0.30^***	0.88^***	1
Privacy concerns	0.62^***	0.63^***	0.70^***	0.73^***	0.68^***	0.69^***	0.65^***	0.16^*	0.75^***	0.77^***	1
Psychological distress	0.39^***	0.41^***	0.44^***	0.45^***	0.41^***	0.45^***	0.44^***	0.05	0.44^***	0.50^***	0.48^***	1
M	5.71	5.6	5.69	5.69	5.72	5.75	5.76	3.97	5.61	5.61	5.59	7.56
SD	0.93	0.93	0.80	0.83	0.88	0.82	0.93	0.43	0.94	0.90	1.00	2.32

p < 0.05; ^** p < 0.01; ^*** p < 0.001.

Table 4.

Hierarchical Regression Analysis Predicting Intention to Use/Continue Using Artificial Intelligence Chatbots

	INTENTION TO USE FOR AI NONUSERS	INTENTION TO CONTINUE USE FOR AI USERS
	MODEL 1	MODEL 2
Performance expectancy	0.60 (0.08)^***	0.46 (0.06)^***	H1a
Effort expectancy	−0.28 (0.11)^*	0.09 (0.09)	H1b
Facilitating conditions	0.12 (0.12)	−0.12 (0.09)	H1c
Price value	0.30 (0.09)^**	0.23 (0.09)^**	H1d
Injunctive norm	−0.090 (0.08)	0.19 (0.08)^*	H1e
Descriptive norm	0.30 (0.06)^***	0.21 (0.06)^**	H1f
Self-stigma	−0.09 (0.09)	0.01 (0.09)	H3a
Social stigma	0.12⁺ (0.07)	0.07 (0.08)	H3b
AI hesitancy	−0.19 (0.07)^*	−0.11 (0.08)	H2a
Privacy concerns	0.02 (0.07)	−0.02 (0.06)	H2b
Psychological distress	0.09 (0.03)^**	−0.0003 (0.02)	H3c
Privacy concerns^* AI hesitancy	−0.02 (0.036)	−0.04 (0.05)	RQ1a
Social stigma^* AI hesitancy	0.02 (0.04)	0.006 (0.04)	RQ1b
Intercept	0.13 (0.70)	−0.004 (0.43)
R-sq	0.672	0.809
adj. R-sq	0.650	0.795
F	3.5	55.7
P	<0.001	<0.001
N	208	185

Unstandardized coefficients (standard errors in parentheses);

+p < 0.1; ^* p < 0.05; ^** p < 0.01; and ^*** p < 0.001.

Among AI nonusers (Model 1), we found that performance expectancy (H1a, b = 0.60, p < 0.001), price value (H1d, b = 0.30, p < 0.01), descriptive norm (H1e, b = 0.30, p < 0.001), and psychological distress (H3c, b = 0.09, p < 0.01) were positively related to intention to adopt AI for mental health. Meanwhile, effort expectancy (H1b, b = −0.28, p < 0.05) and AI hesitancy (H2a, b = −0.19, p < 0.05) were negatively related to intention to adopt AI among nonusers.

Facilitating conditions (H1c, b = 0.12, p > 0.05), self-stigma (H3a, b = −0.09, p > 0.05), social stigma (H3b, b = 0.12, p > 0.05), and privacy concerns (H2b, b = 0.02, p > 0.05) were unrelated to intention to adopt AI among nonusers. We did not find interaction effects between privacy concerns and AI hesitancy (RQ1a, b = −0.02, p > 0.05) and between social stigma and AI hesitancy (RQ1b, b = 0.02, p > 0.05) on the intention to adopt among AI nonusers.

Among AI users (Model 2), performance expectancy (H1a, b = 0.46, p < 0.001), price value (H1d, b = 0.23, p < 0.01), and descriptive norm (H1e, b = 0.21, p < 0.01) were positively related to the intention to adopt AI for mental health. Unlike AI nonusers, personal injunctive norms (H1f, b = 0.19, p < 0.05) were also positively related to the intention to continue using AI chatbots.

However, effort expectancy (H1b, b = 0.09, p > 0.05), AI hesitancy (H2a, b = −0.11, p > 0.05), and psychological distress (RQ1c, b = −0.0003, p > 0.05) were unrelated to the intention to continue using AI chatbots. The interactions between privacy concerns and AI hesitancy (RQ1a, b = −0.04, p > 0.05) and social stigma and AI hesitancy (RQ1b, b = 0.006, p > 0.05) were not significant predictors of intention to continue using AI chatbots among AI users, indicating the lack of moderation among the three predictors for AI users as well.

Discussion

AI chatbots have become increasingly popular in health care.⁶² To examine predictors of both adoption and continued use of AI chatbots for mental health, an online survey was conducted among users and nonusers of AI chatbots who had depression and anxiety symptoms.

We tested critical predictors in the related domains, including mental health help-seeking behavior (i.e., social stigma and self-stigma), AI perception (i.e., AI hesitancy and privacy concern), and technology adoption (i.e., UTAUT2). Findings indicated that UTAUT2 was a viable model to predict the adoption and continued use of AI chatbots for mental health.

One unexpected finding was that effort expectancy negatively predicted adoption intention for people who have never used AI chatbots. A possible explanation is that the expected ease of using mental health chatbots might reduce people's perceived expertise or functionality of AI chatbots, especially for people who have never interacted with or experienced AI chatbots.

In other words, a high degree of effort expectancy might reflect the expectation that mental health chatbots would be limited in their ability to assist users in overcoming mental health issues, as they would only be able to perform simple and basic functions. This expectation may stem from nonusers' experiences with other types of AI chatbots, such as customer service chatbots, which often have limited functions.

However, unlike many simple tasks usually carried out by chatbots, mental health counseling is a complex task that can be challenging even for humans. Therefore, an easy solution using AI chatbots with minimal individual effort may not be perceived as credible and thus reduces individuals' adoption intention. For individuals who have already used AI chatbots, effort expectancy was not a significant predictor of continued use, suggesting that existing users may value other aspects of AI chatbots for continuance.

Social influence was operationalized separately as descriptive and injunctive norms.²⁶ We found that both the injunctive and descriptive norms predicted intention to continue to use AI chatbots. For nonusers, only the descriptive norm predicted adoption intention. The findings suggest that observing other people using AI chatbots is more critical in the adoption decision than recommendations by others to use AI chatbots for mental health issues. This finding implies that it is more important to demonstrate that many other people with similar problems are using AI chatbots for mental health rather than just encouraging adoption.

None of the concepts we imported from mental health help-seeking literature were found to be significant predictors of adoption or continued use, except that psychological distress predicted adoption intention. Although stigma has been consistently found to be an essential factor preventing individuals from seeking mental health help in an offline setting,⁴⁹ the findings suggest that concerns about social or personal stigma may not hinder people from trying or continuing to use AI chatbots, possibly because of the lack of judgment and anonymity provided by AI chatbots.

Consistent with our hypothesis, AI hesitancy was found to lower adoption intention. This finding suggests that in promoting AI-based tools for mental health, it is essential to explain to users how AI chatbots work and why they can be effective in combating their hesitancy. Surprisingly, privacy concern was not related to adoption or continuance intention. Prior research found that privacy was critical in adopting technology-based mental health tools.^63,64

In the case of AI chatbots, it seems that for users and nonusers, the fact that the AI chatbot is not an actual human therapist or counselor might have relieved the privacy concern. However, privacy concern was higher among users than nonusers. Whether the users truly understand the underlying data structure for their privacy concern needs to be further explored.

LIMITATIONS

There are several limitations to the current study. First, given that our sample is a convenience sample, our results are not representative. Second, we did not find any significant moderation, which could be due to our limited sample size rather than a lack of interaction effect. Third, our study was conducted before the release of ChatGPT.

Notably, more people have tried ChatGPT for various daily tasks, such as writing e-mails, summarizing long texts, generating business ideas, or creating software.⁶⁵ Their individual experiences with ChatGPT could sway their opinion on adopting AI chatbots for mental health concerns, such that a positive encounter may boost their confidence about AI chatbots' capabilities to provide solid mental health advice. At the same time, a negative one may raise doubt about its reliability.

Experts also reported heightened concerns about issues such as the privacy of personal data⁶⁶ and algorithmic bias of ChatGPT.⁶⁷ These factors could all complicate individuals' attitudes toward chatbots. Nevertheless, our data contribute insights regarding the use of AI chatbots before ChatGPT, which also serve as a valuable starting point for understanding the evolution of the public's acceptance of AI chatbots.

Conclusions

The coronavirus disease 2019 (COVID-19) pandemic has worsened the mental health of millions of Americans with an overburdened health care system that cannot meet the overwhelming demands.⁶⁸ AI chatbots provide a valuable tool for alleviating mental health concerns, especially those remediable by CBT.

As AI chatbots have the potential to provide accessible, affordable, efficient, and personalized mental health support experience,²³ findings of the study could help drive interventions that target the significant factors for predicting the adoption and continued use of AI chatbots, such as performance expectancy, price value, personal description norm, and AI hesitancy.

Footnotes

Authors' Contributions

L.L. was involved in conceptualization, methodology, formal analysis, investigation, writing—original draft, writing—review and editing, project administration, and funding acquisition. W.P. was involved in conceptualization, methodology, writing—original draft, writing—review and editing, and supervision. M.R. was involved in conceptualization, methodology, formal analysis, writing—review and editing, and funding acquisition.

Disclosure Statement

No competing financial interests exist.

Funding Information

This study is funded by the Charles J. Strosacker Foundation Research Fund for Health and Risk Communication.

References

Anonymous. Any Anxiety Disorder. n.d. Available from: https://www.nimh.nih.gov/health/statistics/any-anxiety-disorder [Last accessed: February 28, 2023 ].

Ettman

, Abdalla

, Cohen

, et al. Prevalence of depression symptoms in US adults before and during the COVID-19 pandemic. JAMA Netw Open, 2020; 3(9):e2019686; doi: 10.1001/jamanetworkopen.2020.19686

Mohr

, Ho

, Duffecy

, et al. Perceived barriers to psychological treatments and their relationship to depression. J Clin Psychol, 2010; 66(4):394–409; doi: 10.1002/jclp.20659

Mohr

, Hart

, Howard

, et al. Barriers to psychotherapy among depressed and nondepressed primary care patients. Ann Behav Med, 2006; 32(3):254–258; doi: 10.1207/s15324796abm3203_12

Rickwood

, Deane

, Wilson

, et al. Young people's help-seeking for mental health problems. Aust E J Adv Ment Health, 2005; 4(3):218–251; doi: 10.5172/jamh.4.3.218

Gorman

JM.

Comorbid depression and anxiety spectrum disorders. Depress Anxiety, 1996; 4(4):160–168; doi: 10.1002/(SICI)1520-6394(1996)4:4<160::AID-DA2>3.0.CO;2-J

Nutt

, Ballenger

, Sheehan

, et al. Generalized anxiety disorder: comorbidity, comparative biology and treatment. Int J Neuropsychopharmacol, 2002; 5(4):315–325; doi: 10.1017/S1461145702003048

Graham

, Greene

, Kwasny

, et al. Coached mobile app platform for the treatment of depression and anxiety among primary care patients: A randomized clinical trial. JAMA Psychiatry, 2020; 77(9):906; doi: 10.1001/jamapsychiatry.2020.1011

Haley

Problem-Solving Therapy: New Strategies for Effective Family Therapy. Jossey-Bass: San Francisco, California; 1977.

10.

Hayes

, Luoma

, Bond

, et al. Acceptance and commitment therapy: Model, processes and outcomes. Behav Res Ther, 2006; 44(1):1–25; doi: 10.1016/j.brat.2005.06.006

11.

Mohr

, Boudewyn

, Goodkin

, et al. Comparative outcomes for individual cognitive-behavior therapy, supportive-expressive group psychotherapy, and sertraline for the treatment of depression in multiple sclerosis. J Consult Clin Psychol, 2001; 69(6):942–949.

12.

Kohn

, Ali

, Puac-Polanco

, et al. Mental health in the Americas: An overview of the treatment gap. Rev Panam Salud Públic, 2018; 42:e165; doi: 10.26633/RPSP.2018.165

13.

Caldarini

, Jaf

, McGarry

. A literature survey of recent advances in chatbots. Information, 2022; 13(1):41; doi: 10.3390/info13010041

14.

Pham

, Nabizadeh

, Selek

. Artificial intelligence and chatbots in psychiatry. Psychiatr Q, 2022; 93(1):249–253; doi: 10.1007/s11126-022-09973-8

15.

Koulouri

, Macredie

, Olakitan

. Chatbots to support young adults' mental health: An exploratory study of acceptability. ACM Trans Interact Intell Syst, 2022; 12(2):11:1–11:39; doi: 10.1145/3485874

16.

Eliot

People Are Eagerly Consulting Generative AI ChatGPT For Mental Health Advice, Stressing Out AI Ethics And AI Law. 2023. Available from: https://www.forbes.com/sites/lanceeliot/2023/01/01/people-are-eagerly-consulting-generative-ai-chatgpt-for-mental-health-advice-stressing-out-ai-ethics-and-ai-law [Last accessed: March 1, 2023 ].

17.

van Dis

EAM

, Bollen

, Zuidema

, et al. ChatGPT: Five priorities for research. Nature, 2023; 614(7947):224–226; doi: 10.1038/d41586-023-00288-7

18.

Kissinger

, Schmidt

, Huttenlocher

ChatGPT Heralds an Intellectual Revolution—WSJ. Wall Str J, 2023. Available from: https://www.wsj.com/articles/chatgpt-heralds-an-intellectual-revolution-enlightenment-artificial-intelligence-homo-technicus-technology-cognition-morality-philosophy-774331c6

19.

Abd-alrazaq

, Alajlani

, Alalwan

, et al. An overview of the features of chatbots in mental health: A scoping review. Int J Med Inf, 2019; 132:103978; doi: 10.1016/j.ijmedinf.2019.103978

20.

Boucher

, Harake

, Ward

, et al. Artificially intelligent chatbots in digital mental health interventions: A review. Expert Rev Med Devices, 2021; 18(sup1):37–49; doi: 10.1080/17434440.2021.2013200

21.

Fitzpatrick

, Darcy

, Vierhile

. Delivering Cognitive behavior therapy to young adults with symptoms of depression and anxiety using a fully automated conversational agent (Woebot): A randomized controlled trial. JMIR Ment Health, 2017; 4(2):e19; doi: 10.2196/mental.7785

22.

Abd-Alrazaq

, Rababeh

, Alajlani

, et al. Effectiveness and safety of using chatbots to improve mental health: Systematic review and meta-analysis. J Med Internet Res, 2020; 22(7):e16021; doi: 10.2196/16021

23.

Vaidyam

, Wisniewski

, Halamka

, et al. Chatbots and conversational agents in mental health: A review of the psychiatric landscape. Can J Psychiatry, 2019; 64(7):456–464; doi: 10.1177/0706743719828977

24.

Malik

, Ambrose

, Sinha

. Evaluating user feedback for an artificial intelligence–enabled, cognitive behavioral therapy–based mental health app (Wysa): Qualitative thematic analysis. JMIR Hum Factors, 2022; 9(2):e35668; doi: 10.2196/35668

25.

Brown

JEH

, Halpern

. AI chatbots cannot replace human interactions in the pursuit of more inclusive mental healthcare. SSM Ment Health, 2021; 1:100017; doi: 10.1016/j.ssmmh.2021.100017

26.

Yuan

, Ma

, Kanthawala

, et al. Keep using my health apps: discover users' perception of health and fitness apps with the UTAUT2 model. Telemed J E Health, 2015; 21(9):735–741; doi: 10.1089/tmj.2014.0148

27.

Bhattacherjee

Understanding information systems continuance: An expectation-confirmation mode. MIS Q, 2001; 25(3):351–370; doi: 10.2307/3250921

28.

Bhattacherjee

, Lin

C-P

. A unified model of IT continuance: Three complementary perspectives and crossover effects. Eur J Inf Syst, 2015; 24(4):364–373; doi: 10.1057/ejis.2013.36

29.

Lazar

, Koehler

, Tanenbaum

, et al. Why We Use and Abandon Smart Devices. In: Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. UbiComp’15 ACM:. New York, NY; 2015; pp. 635–646; doi: 10.1145/2750858.2804288

30.

Epstein

, Caraway

, Johnston

, et al. Beyond Abandonment to Next Steps: Understanding and Designing for Life After Personal Informatics Tool Use. In: Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems. CHI’16 ACM: New York, NY; 2016; pp. 1109–1113; doi: 10.1145/2858036.2858045

31.

Epstein

, Kang

, Pina

, et al. Reconsidering the Device in the Drawer: Lapses as a Design Opportunity in Personal Informatics. In: Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing. UbiComp ’16 Association for Computing Machinery: New York, NY, USA; 2016; pp. 829–840; doi: 10.1145/2971648.2971656

32.

van Bussel

MJP

, Odekerken–Schröder

, Ou

, et al. Analyzing the determinants to accept a virtual assistant and use cases among cancer patients: A mixed methods study. BMC Health Serv Res, 2022; 22(1):890; doi: 10.1186/s12913-022-08189-7

33.

Henkel

, Linn

, van der Goot

. Understanding the Intention to Use Mental Health Chatbots Among LGBTQIA+ Individuals: Testing and Extending the UTAUT. In: Chatbot Research and Design. ( Følstad

, Araujo

, Papadopoulos

, et al. eds). Lecture Notes in Computer Science Springer International Publishing: Cham; 2023; pp. 83–100; doi: 10.1007/978-3-031-25581-6_6

34.

Gursoy

, Chi

, Lu

, et al. Consumers acceptance of artificially intelligent (AI) device use in service delivery. Int J Inf Manag, 2019; 49:157–169; doi: 10.1016/j.ijinfomgt.2019.03.008

35.

Ashfaq

, Yun

, Yu

, et al. I, Chatbot: Modeling the determinants of users' satisfaction and continuance intention of AI-powered service agents. Telemat Inform, 2020; 54:101473; doi: 10.1016/j.tele.2020.101473

36.

Chang

I-C

, Shih

Y-S

, Kuo

K-M

. Why would you use medical chatbots? Interview and survey. Int J Med Inf, 2022; 165:104827; doi: 10.1016/j.ijmedinf.2022.104827

37.

Terblanche

, Kidd

. Adoption factors and moderating effects of age and gender that influence the intention to use a non-directive reflective coaching chatbot. SAGE Open, 2022; 12(2):21582440221096136; doi: 10.1177/21582440221096136

38.

Prakash

, Das

, Indian Institute of Technology

Kharagpur

. Intelligent conversational agents in mental healthcare services: A thematic analysis of user perceptions. Pac Asia J Assoc Inf Syst, 2020; 12:1–34; doi: 10.17705/1thci.12201

39.

Kohnke

, Cole

, Bush

. Incorporating UTAUT predictors for understanding home care patients' and clinician's acceptance of healthcare telemedicine equipment. J Technol Manag Amp Innov, 2014; 9(2):29–41; doi: 10.4067/S0718-27242014000200003

40.

Cimperman

, Makovec Brenčič

, Trkman

. Analyzing older users' home telehealth services acceptance behavior—Applying an Extended UTAUT model. Int J Med Inf, 2016; 90:22–31; doi: 10.1016/j.ijmedinf.2016.03.002

41.

Aggelidis

, Chatzoglou

. Using a modified technology acceptance model in hospitals. Int J Med Inf, 2009; 78(2):115–126; doi: 10.1016/j.ijmedinf.2008.06.006

42.

Tavares

, Oliveira

. Electronic Health Record Portal Adoption: A cross country analysis. BMC Med Inform Decis Mak, 2017; 17(1):97; doi: 10.1186/s12911-017-0482-9

43.

Damberg

Predicting future use intention of fitness apps among fitness app users in the United Kingdom: The role of health consciousness. Int J Sports Mark Spons, 2021; 23(2):369–384; doi: 10.1108/IJSMS-01-2021-0013

44.

Nadarzynski

, Miles

, Cowie

, et al. Acceptability of artificial intelligence (AI)-led chatbot services in healthcare: A mixed-methods study. Digit Health, 2019; 5:2055207619871808; doi: 10.1177/2055207619871808

45.

Montagni

, Tzourio

, Cousin

, et al. Mental Health-related digital use by university students: A systematic review. Telemed E Health, 2020; 26(2):131–146; doi: 10.1089/tmj.2018.0316

46.

Parker

, Halter

, Karliychuk

, et al. How private is your mental health app data? An empirical study of mental health app privacy policies and practices. Int J Law Psychiatry, 2019; 64:198–204; doi: 10.1016/j.ijlp.2019.04.002

47.

Dang

, Guo

, et al. Privacy concerns about health information disclosure in mobile health: Questionnaire study investigating the moderation effect of social support. JMIR MHealth UHealth, 2021; 9(2):e19594; doi: 10.2196/19594

48.

Mak

WWS

, Poon

CYM

, Pun

LYK

, et al. Meta-analysis of stigma and mental health. Soc Sci Med, 2007; 65(2):245–261; doi: 10.1016/j.socscimed.2007.03.015

49.

Schnyder

, Panczak

, Groth

, et al. Association between mental health-related stigma and active help-seeking: Systematic review and meta-analysis. Br J Psychiatry, 2017; 210(4):261–268; doi: 10.1192/bjp.bp.116.189464

50.

Rickwood

, Braithwaite

. Social-psychological factors affecting help-seeking for emotional problems. Soc Sci Med, 1994; 39(4):563–572; doi: 10.1016/0277-9536(94)90099-X

51.

Carr

‘AI gone mental’: Engagement and ethics in data-driven technology for mental health. J Ment Health, 2020; 29(2):125–130; doi: 10.1080/09638237.2020.1714011

52.

Miles

, West

, Nadarzynski

. Health chatbots acceptability moderated by perceived stigma and severity: A cross-sectional survey. Digit Health, 2021; 7:20552076211063012; doi: 10.1177/20552076211063012

53.

Venkatesh

, Thong

JYL

, Xu

. Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology. MIS Quarterly, 2012; 36(1):157–178; doi: 10.2307/41410412

54.

Park

, Smith

. Distinctiveness and influence of subjective norms, personal descriptive and injunctive norms, and societal descriptive and injunctive norms on behavioral intent: A case of two behaviors critical to organ donation. Hum Commun Res, 2007; 33(2):194–218; doi: 10.1111/j.1468-2958.2007.00296.x

55.

Vogel

, Wade

, Haake

. Measuring the self-stigma associated with seeking psychological help. J Couns Psychol, 2006; 53(3):325–337; doi: 10.1037/0022-0167.53.3.325

56.

Komiya

, Good

, Sherrod

. Emotional openness as a predictor of college students' attitudes toward seeking psychological help. J Couns Psychol, 2000; 47(1):138–143; doi: 10.1037/0022-0167.47.1.138

57.

Smith

, Milberg

, Burke

. Information Privacy: Measuring individuals' concerns about organizational practices. MIS Q, 1996; 20(2):167–196; doi: 10.2307/249477

58.

Buchanan

, Paine

, Joinson

, et al. Development of measures of online privacy concern and protection for use on the Internet. J Am Soc Inf Sci Technol, 2007; 58(2):157–165; doi: 10.1002/asi.20459

59.

Sinkovics

, Stöttinger

, Schlegelmilch

, et al. Reluctance to use technology-related products: Development of a technophobia scale. Thunderbird Int Bus Rev, 2002; 44(4):477–494; doi: 10.1002/tie.10033

60.

Spitzer

, Kroenke

, Williams

JBW

, et al. A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch Intern Med, 2006; 166(10):1092–1097; doi: 10.1001/archinte.166.10.1092

61.

Kroenke

, Spitzer

, Williams

JBW

. The Patient Health Questionnaire-2: Validity of a two-item depression screener. Med Care, 2003; 41(11):1284–1292; doi: 10.1097/01.MLR.0000093487.78664.3C

62.

Beets

, Newman

, Howell

, et al. Surveying public perceptions of artificial intelligence in health care in the United States: Systematic review. J Med Internet Res, 2023; 25(1):e40337; doi: 10.2196/40337

63.

Martinez-Martin

, Kreitmair

. Ethical issues for direct-to-consumer digital psychotherapy apps: Addressing accountability, data protection, and consent. JMIR Ment Health, 2018; 5(2):e9423; doi: 10.2196/mental.9423

64.

Huckvale

, Nicholas

, Torous

, et al. Smartphone apps for the treatment of mental health conditions: Status and considerations. Curr Opin Psychol, 2020; 36:65–70; doi: 10.1016/j.copsyc.2020.04.008

65.

Mollick

ChatGPT Is a Tipping Point for AI. Harv Bus Rev, 2022. Available from: https://hbr.org/2022/12/chatgpt-is-a-tipping-point-for-ai

66.

Burgess

ChatGPT Has a Big Privacy Problem. Wired n.d. Available from: https://www.wired.com/story/italy-ban-chatgpt-privacy-gdpr/

67.

Rozado

The political biases of ChatGPT. Soc Sci, 2023; 12(3):148; doi: 10.3390/socsci12030148

68.

Nielsen

, Levkovich

. COVID-19 and mental health in America: Crisis and opportunity?. Fam Syst Health J Collab Fam Healthc, 2020; 38(4):482–485; doi: 10.1037/fsh0000577