The World Health Organization (WHO) describes mental health as a state of well-being where an individual recognizes their abilities, manages everyday stress, works effectively, and contributes to their community (1). Mental health disorders include various issues such as depression, anxiety, addiction, bipolar disorder, and other disorders, which significantly affect a person’s daily life, relationships, and physical health (2). In recent years there has been rapid uptake and integration of Artificial Intelligence (AI) in mental healthcare with the goal of improving patient care through greater efficiency in processes, prediction, and resources to individualize care (3, 4). While there is no standard definition for AI, the American Psychological Association defines it as ‘a subdiscipline of computer science that aims to produce programs that simulate human intelligence’ (5).

In practice there are different types of AI including machine learning – neural network and deep learning; and natural language processing (NLP); rule-based expert systems; robotic process automation and physical robots (6, 7). Research in artificial intelligence (AI) has primarily focused on components such as learning, reasoning, problem-solving, decision making, creativity, perception, independence and language use (8–10). Benefits of AI include improved efficiency, accuracy, and productivity (11). AI has been extensively used in health care system (12), including mental health systems (3, 13, 14). (6, 15–21) Attitudes toward AI in healthcare are often positive (22–24), with professionals noting the potential of AI to support medical diagnosis, decision-making, drug discovery, patient experience, data management, and robotic surgery (25, 26). However, an important distinction is the perspective among many healthcare professionals is that AI should serve as a partner not a replacement as a way of maximizing the benefits of AI and the expertise of the clinician (27). Other factors raised by healthcare professionals reflecting on the integration of into healthcare relate to job security, ethical and moral dilemmas, impact on quality of care including patient-provider relationships, training needs, and legal implications and regulatory standards (27). AI-driven advancements can improve personalized, precise, and predictive care, while challenges to integration include data access, domain expertise, public trust, ethical concerns, cybersecurity, and language barriers (12, 13, 22). Additionally, effective AI implementation depends on robust evaluation, bias mitigation, generalizability, and interpretability to maximize patient benefits (28). However, limited digital literacy remains a barrier, with studies showing that nearly half (48.2%) of healthcare professionals exhibit poor digital literacy, which may limit use of AI in supporting clinical practice (29).

Out of 251 study participants, 56.6% were females, about 50% were in age group of 31 to 40 years, 45% were married. Professionally, 55.4% were nurses, and 83.1% were in a clinical role. Work experience varied between 5–10 years (27.5%), and 11–20 years (35.9%) ( Table 1 ).

Use or familiarity with AI varied across the sample. While the majority claimed they knew what AI was (85.7%), most had not received any training in its use (72.9%). Notably, while 66.5% stated they were not using AI in their clinical practice, and 24.3% stated they were using it, 9.2% indicated they were unsure ( Table 2 ).

Participant responses to the SHAIP questionnaire highlight important complexities within their attitudes towards AI ( Table 3 ). For example, 61.8% indicated ‘agree’ to the item ‘I believe that AI can improve population health outcomes’, 53.8% agreed that AI would change their healthcare role, and 55% agreed AI may take over portions of their healthcare role. However, responses to preparedness items suggest contradictory perceptions among participants. Only 20.3% agreed they had been adequately trained to use AI within their specialty role, affirmed by 32.7% and 25.1% stating unsure and disagree respectively, yet 48.2 agreed that the healthcare professionals are prepared for the introduction of AI technology.

The mean (Sd.), of the two factors under exploration (Professional Impact of AI, Preparedness of AI), are 21.76(4.26), and 13.52(2.88) respectively ( Figure 1 ). The comparison of mean values of the two factors of the SHAIP questionnaire in relation to the demographic, professional and awareness items on AI show statistically significant differences ( Table 4 ). These were found within the variable of ‘specialty’ and the personal profile characteristic ‘do you know what AI is?’ for both factors of professional impact of AI and preparedness of AI. The variable ‘age’ however, was also found to be significant for differences found within the factor of professional impact of AI. The mean values of professional impact of AI are significantly different across the 4 age groups (p=0.003). The mean values of participants are significantly higher in those aged <20, 31–40 and > 40 years when compared with the participants who are aged 21–30 years.

The mean values of professional impact of AI are significantly higher in participants who were nurses, nurse interns, psychologists, and sociologists when compared with the mean values of participants who were psychiatrists and of other specialties (p=0.008) ( Figure 2 ). The post-hoc test indicates no significant difference in the mean values of pairs of nurses, nurse interns, psychologists and sociologists. The mean value of professional impact of AI is significantly higher among participants who had responded positively (yes) to the statement ‘do you know what AI is?’ when compared with the participants who had responded negatively (No) (p<0.0001). No statistically significant difference was observed for the mean values of professional impact of AI in relation to the other variables ( Table 4 ).

For the mean values of preparedness of AI, we observed statistically significantly higher difference in the mean values among the participants who were nurses, nurse interns, psychologists, and other specialties when compared with the mean values of participants who were psychiatrists and sociologists (p=0.001) ( Figure 3 ). The post-hoc test indicates no significant difference in the mean values of pairs of nurses, nurse interns, psychologists and other specialties. The mean value of preparedness of AI are significantly higher among participants who had responded positively (yes) to the statement ‘do you know what AI?’ when compared with the participants who had responded negatively (No) (p=0.029). No statistically significant difference was observed for the mean values of preparedness of AI in relation to the other variables ( Table 4 ).

This cross-sectional study aimed to explore perceptions of AI among a diverse range of mental healthcare professionals in Saudi Arabia. The study sample consisted predominantly of individuals aged 31–40 years (49.8%), with a significant representation of females (56.6%). A large proportion of participants were married (45.0%), compared to single (39.4%). Professionally, nurses constituted the majority (55.4%), followed by psychologists (13.5%) and nurse interns (9.6%). Most participants worked in clinical roles (83.1%), with a smaller percentage in administrative positions (16.9%). Over a third of participants had 11–20 years of clinical experience (35.9%), while a significant portion had less than 5 years (31.5%). These demographic trends highlight a sample with diverse professional backgrounds and experience levels, which may influence their perceptions and preparedness for AI integration in healthcare. Notably, the underrepresentation of certain specialties, such as psychiatrists (8.0%) and sociologists (6.4%), may limit the generalizability of findings across all healthcare professions.

The findings of this study shed light on three key elements connected to AI preparedness and perceived impact among mental health professionals in Saudi Arabia. First, we found there was an overarching general awareness of AI, but a lack of training supporting its use. This suggests a potential risk for poor or unsafe practices if AI interventions are implemented without establishment of institutional infrastructure to provide training and support. The second area pertains to the perceived professional Impact of AI. Overall, participants were largely positive their perceptions of the role of AI in mental healthcare. However, while about half believed AI would alter their clinical role and potentially take on aspects of their responsibilities, the survey did not capture if they felt this change would also be a positive move – an area requiring further investigation. The third element builds on this aspect, with notable differences in perceptions and preparedness between different professional roles and demographics (explored further below). Nurses, psychologists, and sociologists reported a greater level of perceived impact and preparedness than psychiatrists and some other professionals, associated with higher levels of knowledge. This reinforces the need for training to accompany intervention development if AI is to be integrated to greater effect in the clinical setting. It further suggests a relationship between familiarity and attitude.

Age revealed complex relationships between demographic factors and AI perception. Age emerged as a significant predictor of professional impact but not preparedness, adding nuance to existing literature. While some studies found younger age groups (18–42 years) more receptive to AI mental health tools (38), others reported no significant age-related differences in AI perception and attitudes (24, 39, 40). (41) The age-related variation in our study appears more closely tied to professional impact rather than general preparedness, suggesting that age influences how professionals view AI’s effect on their work rather than their readiness to adopt it. This is further supported by Brinker et al. (2019) who found that diagnostic performance patterns vary with age—younger clinicians showing higher sensitivity and older clinicians better specificity—indicating that AI tools may need to be tailored to different experience levels (47).

Gender showed no significant relationship with AI perception in our study, aligning with several recent studies (39, 40). However, this finding exists within a complex landscape of contradictory evidence. Ofosu-Ampong (2023) found men perceived AI-based learning tools more positively, attributing this to traditional gender roles and socialization patterns (42). Conversely, Fritsch et al. (2022) reported women showed more favorable attitudes toward AI in healthcare (24). Gender differences were also observed in diagnostic performance, with female clinicians showing higher sensitivity and male clinicians greater specificity (47). These contradictory findings suggest the presence of unmeasured moderating variables that warrant further investigation. Our study proved marital status to be a non-significant factor in AI perception, though this demographic variable has received limited attention in previous research.

Professional specialization emerged as a crucial factor in shaping AI perceptions. Psychologists reported the highest professional impact scores, while nurses demonstrated the highest preparedness levels. This finding is particularly noteworthy as previous research often excluded these key healthcare professionals (39). Within mental health specialties, those practicing cognitive behavioral approaches showed more positive attitudes toward AI (43). However, the adoption landscape remains complex, particularly in psychiatry, where evidence gaps regarding AI’s benefits and limitations make implementation decisions challenging (16). This specialty-specific variation in AI perception suggests the need for tailored implementation strategies that consider each profession’s unique needs and concerns. However, our study found no relationship between field of specialty, such as administrative and clinical roles, and AI perception. Participants from both fields exhibited no differences in AI perception across both factors. Years of experience showed no significant relationship with AI perception, consistent with previous findings (39, 40). However, some nuanced differences emerged: senior physicians were reported to be less familiar with AI (40), and more experienced doctors tended to prioritize human expertise over AI in diagnostics (44).

In our analysis, the current use of AI in mental health care settings did not reveal a relationship with AI perception in either professional impact and AI preparedness. We observed no connection between professional’s training in AI and AI perception. Interestingly, professionals who received training and professionals who had not, had similar levels of perceptions towards AI. While in the literature, digital literacy and training in technology emerged as key factors influencing healthcare professionals’ attitudes toward AI adoption (29), in our work, the relationship between professional’s knowledge and AI perception proved significant. Specifically, that those familiar with AI showed a more positive perception. This aligns with findings on technology familiarity’s importance (42), though again, wider literature is not consistent on this relationship (39).

From a cultural perspective, our findings among mental healthcare professionals are reflected among other healthcare specialties in the region. For example, a cross-sectional study in Jeddah, Saudi Arabia found healthcare workers, including mental health professionals, demonstrated substantial awareness of AI, but limited AI experience and training (48). Similarly, a study from the United Arab Emirates pointed to positive attitudes towards the role of AI in certain administrative clinical processes, with limited training and education opportunities (49). More globally, perceptions towards the potential role of AI include questions about its ability or appropriateness to replace human connection and empathy in mental healthcare, along with concerns about ethical elements (50). However, a systematic review about the future role of AI in Saudi healthcare more generally, suggests that its potential benefits in improving care processes mean the work to ensure ethical application is a worthwhile endeavor (51). Within this wider context, the findings from the present study suggest that in Saudi Arabian mental healthcare, gaps in level of preparedness can be supplemented by rigorous training across professions, a need for more work to ensure safe and ethical integration of AI to improve patient care processes, and support and partnership with mental healthcare professionals in understanding the potential evolution of their roles and responsibilities.

Our findings indicate that there are relationships between the demographic variables of healthcare professionals and perceptions and use of AI in practice. Specialization and AI knowledge significantly influence perceptions among mental health professionals. Psychologists reported the highest professional impact, while nurses showed the greatest preparedness. Psychiatrists and sociologists exhibited lower preparedness scores, highlighting potential gaps. Participants familiar with AI had more positive perceptions, emphasizing the role of knowledge. Age affected professional impact but not preparedness. Gender, marital status, experience, and training showed no significant relationships with AI perceptions. These results underscore the need for tailored strategies to address specialty-specific needs and enhance AI literacy.