Assessing Perceptions and Exposure to Artificial Intelligence Among Medical Educators and Clinicians in Saudi Arabia: A Cross-Sectional Study

Summary

This cross-sectional study explored perceptions of artificial intelligence among medical educators and clinicians in Saudi Arabia. Although strong interest was reported, formal AI training was limited, and reliance on informal sources was common, highlighting the potential need for structured AI education.

Abstract

Artificial intelligence (AI) is transforming healthcare through advances in diagnostics and decision-making, yet most studies have focused on medical students’ perceptions, with limited attention to medical educators and practicing clinicians. Understanding these perspectives is critical for developing curricula that prepare future physicians for AI-enhanced practice. This study investigated AI awareness, exposure, and perceived impact among medical educators, interns, residents, specialists, and consultants in Saudi Arabia. A cross-sectional study was conducted using a validated online survey distributed to several Saudi medical schools and hospitals. The survey collected demographic data, prior AI education, self-rated AI knowledge, and interest in AI. The participants also rated their ability to list clinically relevant AI applications, assessed AI’s expected influence on various specialties, and indicated whether training in AI concepts would be beneficial. Group differences were analyzed using chi-square tests, ANOVA, and logistic regression, controlling for age and sex. Among the 229 respondents (60 medical educators, 78 interns/residents, and 91 specialists/consultants), medical educators were more likely to report the ability to list AI benefits (63.3% vs. 38.5% and 39.6%) and to express ethical concerns about AI use in healthcare (76.6% vs. 51.3% and 51.6%) compared to the other groups. Only 32.3% reported receiving formal AI education, with most gaining knowledge through informal sources such as the media. Radiology and pathology were identified as the specialties most likely to be transformed by AI. Overall, perceptions of AI among medical educators and clinicians revealed a strong interest but limited formal training. Despite enthusiasm across all groups, the participants relied on informal sources and lacked foundational knowledge on AI. These findings highlight the potential importance of embedding structured AI education into medical curricula and residency training programs to support preparedness for AI-integrated clinical care.

Introduction

The term “artificial intelligence” (AI) was first introduced at the Dartmouth Conference in 1956 and is now globally regarded as a top technology¹. AI broadly refers to technologies capable of mimicking human intelligence² and is commonly categorized into four domains: natural language processing for speech and text recognition and generation; computer vision for interpreting images; machine learning based on neural networks; and knowledge-based systems that rely on semantic technologies or cognitive modeling³.

One of the most rapidly evolving applications of AI is in healthcare. AI has been described as having the potential to revolutionize healthcare through its integration into clinical practice⁴. By enhancing diagnostic accuracy and optimizing treatment decisions, AI can also transform the way physicians deliver care⁵. For these applications to be effectively implemented, healthcare professionals must possess adequate knowledge and awareness of AI technologies. Its successful application has been documented across multiple specialties, including radiology, pathology, ophthalmology, oncology, and dentistry⁶. In medical education, AI has also supported learning through virtual patient simulations with the aim of developing communication and clinical skills⁷.

Several studies have assessed medical students’ perceptions of AI in healthcare and education^5,8,9,10. These studies have suggested growing interest among students but also highlighted educational gaps. However, only one study has specifically examined students’ exposure to formal AI education⁸. Similarly, multiple studies have explored Saudi practicing clinicians’ perceptions of AI^11,12,13,14, but none have assessed exposure to formal AI education. There remains a notable lack of research specifically focused on medical educators and how they perceive AI and its role in medical education.

To address this gap, this study assessed the current knowledge, exposure, perceptions, and information sources related to AI among medical educators and practicing clinicians in Saudi Arabia, as well as their views on its influence across medical specialties. We conducted a cross-sectional survey to evaluate these domains within this population. The findings could help guide future efforts to integrate AI-related competencies into medical education and residency training programs^15,16,17. To our knowledge, the present study is one of the first in Saudi Arabia to simultaneously examine medical educators’ and practicing clinicians’ exposure to formal AI education, as well as their perceptions and self-reported knowledge of AI.

Protocol

Study modeling
Ethical approval for this study was obtained from the Institutional Review Board of King Abdulaziz University (Reference No. 500-23). Eligible participants included medical professionals aged 18 years or older (medical educators, interns, residents, specialists, and consultants) practicing in Saudi Arabia. Individuals who were not medical graduates or were not actively practicing were excluded. All participants provided informed consent prior to data collection, and their participation was entirely voluntary. An electronic informed consent statement was presented on the first page of the survey, and the participants were required to indicate consent before proceeding. The participants were informed that they could withdraw at any time prior to survey submission without consequences. To maintain confidentiality, no personally identifiable information was collected, and all collected data were de-identified and securely stored.

Data collection tools
Data were collected via a previously validated survey delivered online via Google Forms. Participants were recruited through convenience sampling by posting recruitment materials without a targeted outreach list. The recruitment materials containing the survey link were distributed via electronic advertisements, institutional emails, and online messaging platforms shared with healthcare professionals in Saudi medical schools and hospitals. As such, no formal response rate could be calculated.

The first section of the questionnaire collected demographic information, including date of birth, sex, role in the medical field, specialty, and—for residents—year of training. The second section focused on the participants’ previous exposure to AI. They were asked whether they had received formal education in AI (such as college- or university-based courses). This item was intentionally broad and relied on participants’ self-interpretation of what constituted ‘formal’ AI education. The participants were also asked to identify their sources of AI exposure.

To assess participants’ perceptions and understanding of AI, we used a previously validated survey originally designed to evaluate the perceived impact of AI use in medicine among medical students⁸. The survey included items on AI’s influence on different medical specialties. The participants also rated their self-perceived understanding of foundational AI concepts, including convolutional neural networks (CNNs), cross-validation, receiver operating characteristic (ROC) curves, and area under the curve (AUC). This approach allowed us to explore the gap between participants’ interest in AI and their level of formal preparedness.

Survey responses were automatically recorded within Google Forms and exported to Microsoft Excel for data organization. The data were screened for completeness prior to statistical analysis. De-identified datasets were encrypted and stored on a password-protected drive that was only accessible to members of the research team.

Statistical analysis
The participants were categorized as medical educators, interns/residents, or specialists/consultants. All Likert-scale questions were categorized as either agreement (agree or strongly agree) or non-agreement (neutral, disagree, or strongly disagree) for group comparisons.

Descriptive statistics included the mean and standard deviation of continuous variables and frequencies with proportions for categorical variables. Demographic variables across groups were compared using ANOVA for continuous variables and Chi-square testing for categorical variables. The survey responses (agreement or non-agreement) of the groups were compared using logistic regression modeling (agreement or non-agreement as the outcome) with post hoc testing. All testing applied the Bonferroni correction to adjust for multiple comparisons between groups. Associations were tested against continuous variables (age) using linear multiple regression modeling, and against categorical variables (having had or not had formal education in AI) using logistic multiple regression. All regression models controlled for the participants’ age and sex.

Multicollinearity was tested by calculating the variance inflation factor (VIF) and was found to be negligible in all models. Linear regression model residuals were confirmed to be normally distributed using the Kolmogorov–Smirnov test. Statistical analyses were completed in R version 4.3.1, with all modeling using the base stats package, except for the multicollinearity assessments, which used the usdm package. All significance testing used a threshold of α = 0.05. Final model outputs were exported and used to generate the analyses, tables, and figures presented in the Results section.

Results

Sample
This cross-sectional survey presents observational comparisons among naturally occurring participant groups, consistent with the study's exploratory nature. The sample comprised 229 participants (60 medical educators, 78 interns/residents, and 91 specialists/consultants). The groups did not significantly differ in terms of sex (47.6% male, p=0.088 between groups) or having had a formal AI education (32.3% with training, p=0.067 between groups). Interns and residents were significantly younger ...

Discussion

To our knowledge, this is one of the first studies in Saudi Arabia to examine how medical educators and practicing clinicians perceive AI. The strong reported interest aligns with the broader adoption of AI in healthcare, which has been documented in both clinical and educational settings¹⁸. This trend has been met with positive acceptance among medical educators, especially in the context of medical education, where tools such as ChatGPT have been recognized for their usefulness in tasks like sum...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Google Forms	Google LLC	N/A	Used to distribute and collect responses for the online survey
Microsoft Excel	Microsoft Corporation	N/A	Used for data organization and export prior to analysis
PubMed Database	U.S. National Library of Medicine	N/A	Used for literature review and retrieval of references
Questionnaire	Liu D, Sawyer J, Luna A, Aoun J, Wang J, Boachie L, Halabi S, Joe B Perceptions of US Medical Students on Artificial Intelligence in Medicine: Mixed Methods Survey Study JMIR Med Educ 2022;8(4):e38325 URL: https://mededu.jmir.org/2022/4/e38325 DOI: 10.2196/38325	N/A	Validated questionnaire assessing perceptions of AI in medicine, adapted from a previously published study in JMIR Medical Education (2022)
R (Statistical Software)	R Foundation for Statistical Computing	N/A	Used for statistical analysis and data visualization