Skip to main content
NCBI home page
Advances in Medical Education and Practice logoLink to Advances in Medical Education and Practice
. 2026 Sep 10;17:633577. doi: 10.2147/AMEP.S633577

Artificial Intelligence in Medical Education: Perceptions of Romanian Medical Students in a Cross-Sectional Study

Oana Raluca Radbea 1,2,3, Dumitru Sutoi 2,4,5,6,✉, Bogdan Chiu 2,✉, Daian Ionel Popa 1,2,5, Ionut Dudau 1, George Marin 2, Anda Nicoleta Ciontos 2, Vlad Mulcutan-Chis 2, Maria Sutoi 7, Raul Chioibaș 2,8, Ovidiu Alexandru Mederle 2,4,5,6
PMCID: PMC13575613  PMID: 42746198

Abstract

Background

The rapid integration of artificial intelligence (AI) into medical education has transformed learning environments and educational strategies. Understanding medical students’ perceptions regarding the educational utility, clinical applicability, and ethical implications of AI tools is essential for guiding their responsible integration into modern medical curricula.

Methods

A cross-sectional online survey was conducted among medical students across all six study years at Victor Babeș University of Medicine and Pharmacy Timișoara in May 2026 using a modified, validated instrument. Questionnaire domains demonstrated good-to-excellent internal consistency (Cronbach’s α=0.80–0.94). Data from 374 valid respondents were evaluated using non-parametric Kruskal–Wallis tests with Holm–Bonferroni-adjusted Dunn’s post-hoc contrasts and multivariable ordinary least squares linear regression.

Results

Overall, 95.98% (n=359) of students reported prior AI use, and 62.56% (n=234) indicated frequent or daily utilization. Students reported the highest agreement for AI saving time during information retrieval (83.42%) and assisting in understanding medical concepts (74.33%). Cross-sectional comparisons identified significant differences across study years for five items (ε2=0.031–0.052), including higher perceived time savings (padj=0.007) and effort reduction (padj=0.040) among 6th-year compared to 3rd-year students. In multivariable linear regression adjusting for demographic and academic covariates, a positive attitude toward AI was a strong independent predictor of perceived clinical utility (B=0.899, SE=0.042, β=0.755, p<0.001, R2=0.577).

Conclusion

Romanian medical students demonstrated high familiarity with AI applications and favorable perceptions of their educational and clinical utility, with positive attitudes strongly predicting perceived clinical usefulness. Cross-sectional variations emerged across specific items and study cohorts. Persistent concerns regarding accuracy, reliability, and academic overdependence highlight the need for structured AI literacy curricula and clear institutional guidelines.

Keywords: artificial intelligence, AI applications, medical students, medical education, digital learning, higher education

Introduction

Medical education has undergone substantial digital transformation in recent years, with e-learning platforms, simulation-based training, interactive resources, and artificial intelligence (AI)-based applications increasingly complementing conventional teaching methods.1,2 The rapid development of generative AI and large language models has further expanded these possibilities by providing medical students with accessible tools for personalized explanations, rapid feedback, information retrieval, summarization of scientific content, examination preparation, and self-directed learning.3–7 These technologies are therefore increasingly being considered as potential educational resources capable of supporting both theoretical learning and clinically oriented training.

Within medical education, AI-assisted tools have been explored as potential resources for supporting learning, access to medical information, and clinically oriented educational activities.8–15 Proposed applications include case-based explanations, information retrieval, differential diagnosis exercises, clinical reasoning exercises, simulation-based learning, objective structured clinical examination preparation, and formative assessment.11–15 Importantly, the perceived usefulness and proposed educational applications of these tools should not be interpreted as evidence of objectively demonstrated improvements in learning outcomes, clinical reasoning, or clinical performance. Generative AI systems may also produce inaccurate or fabricated information, introduce algorithmic bias, and provide responses with limited transparency.16–18 Concerns have further been raised regarding academic integrity, excessive dependence on AI-assisted learning, potential effects on independent critical thinking, data privacy, and the need for appropriate institutional oversight.19,20 Consequently, the effective integration of AI into medical education requires not only technological accessibility but also an understanding of how students perceive educational value, clinical applicability, reliability, and ethical implications.

Despite increasing international research on medical students’ attitudes toward AI, evidence remains geographically uneven. Most available studies have been conducted in North America, Western Europe, or Asia, while data from Eastern European medical education settings remain comparatively limited.21,22 Little is known about how Romanian medical students perceive the educational and clinical usefulness of AI applications, how they evaluate their limitations and ethical implications, and how these perceptions are expressed within the specific context of undergraduate medical training in Romania. Addressing this gap is important because students’ experiences and attitudes may help inform the development of locally appropriate AI literacy initiatives, curricular strategies, and institutional guidance.

Therefore, the primary objective of the present study was to evaluate Romanian medical students’ perceptions and attitudes regarding the use of AI applications in medical education. Secondary objectives were to assess their perceived educational and clinical utility, including students’ perceptions of the potential role of AI in supporting learning efficiency and clinical reasoning; to examine perceived limitations, risks, and ethical concerns associated with AI use; and to explore students’ views regarding the future integration of AI technologies into medical education and clinical training.

Materials and Methods

Study Design and Participants

A cross-sectional questionnaire-based study was conducted among medical students enrolled at Victor Babeș University of Medicine and Pharmacy, Timișoara, in May 2026. This study aimed to evaluate Romanian medical students’ perceptions, attitudes, and experiences regarding the use of artificial intelligence in medical education.

Participation in the study was voluntary and anonymous. The questionnaire was distributed online via Google Forms sent in each years WhatsApp group, where all students from a certain year are present. Besides this, the WhatsApp link was sent to each series representatives to forward the link to each series of students of each year. A total of 401 responses were initially collected. The inclusion criteria were enrollment as a medical student at the Faculty of Medicine of the „Victor Babeș” University of Medicine and Pharmacy Timișoara, regardless of the year of study, and voluntary agreement to participate in the study through informed consent. The exclusion criteria were refusal to provide informed consent and incomplete or incorrectly completed questionnaires.

Among the 401 responses collected, 2 participants did not provide informed consent, and 25 questionnaires were excluded because they were incomplete. Consequently, 374 valid responses were included in the final statistical analysis. Based on an eligible source population of approximately 3,000 medical students enrolled across all study years at Victor Babeș University of Medicine and Pharmacy Timișoara, a minimum sample size of 341 participants (using Cochran’s sample size formula for finite populations) was required to achieve a 95% confidence level with a 5% margin of error. Net response rate was calculated to be approximately 12.47%. No formal statistical power calculation was performed prior to study initiation. The participant selection process is illustrated in Figure 1.

Figure 1.

A flowchart illustrating the process of questionnaire response selection and analysis. The flowchart outlines the process of handling questionnaire responses. It begins with ′Questionnaire Distributed Responses Received′ with n equals 401 under the identification stage. In the selection stage, ′Responses Assessed for Eligibility′ with n equals 401 are shown, leading to two exclusions: ′Excluded due to missing informed consent′ with n equals 2 and ′Excluded: Incomplete Responses′ with n equals 25. The eligibility stage shows ′Responses with Approved Informed Consent′ with n equals 399. The inclusion stage presents ′Complete Responses Included in the Analysis′ with n equals 374. Finally, the analysis stage concludes with ′Data Analysis and Interpretation of Results′.

Open in a new tab

Study flowchart.

Questionnaire Description

The survey instrument was developed using a modified questionnaire adapted from a previously published study investigating perceptions of artificial intelligence in medical education.23 The questionnaire was revised and adjusted to better reflect the educational and sociocultural context of Romanian medical students. Out of the total questions used in the study mentioned, only the ones regarding family status (“Family status” and “Family income”) to ensure participant privacy and a more honest response. Besides, the “=” sign has been added to the “>21 years old” category to include the students who are 21 years old. In addition to the initial block of questions, another section has been added, entitled “Clinical applicability of AI usage”. In this part, 10 original Likert-type questions have been created, to evaluate how AI is used by students in a clinical setting. The main focus of them was as before, on the understanding of concepts, critical thinking and educational purpose of AI. This final questionnaire has been evaluated by 10 students from higher years of study and then evaluated by two professors with experience in the medical education domain to asses lingual appropriation and quality of question to asses the research hypothesis. After deciding on the block of questions, they have been integrated into Google Forms after undergoing direct Romanian translation from the original questionnaire. No applications have been used to translate, only direct translation by the research team has been done and then evaluated for correctness by two professors mentioned before.

The final questionnaire consisted of multiple sections designed to assess students’ demographic characteristics, familiarity with artificial intelligence technologies, patterns of AI utilization in academic activities, perceived advantages and limitations of AI in medical education, ethical concerns, and attitudes regarding the future integration of AI into healthcare training. The final questionnaire was composed of seven sections: “Informed Consent”, “Demographical information”, “Frequency of AI usage”, “Perception regarding AI of medical students”, “Attitude towards AI for medical students”, “Practical usage of AI apps”, “Clinical applicability of AI apps”. The questionnaire included both closed-ended and Likert-scale items (from 1 to 5), where 1 represented complete disagreement with the statement and 5 represented complete agreement, with 3 being the middle point. The demographic variables included age, gender, and year of study, area of residence, actual living arrangements, and the average grade for the prior university year.

Data Collection

Data collection was performed electronically throughout May 2026 using the Google Forms platform. Participants accessed the questionnaire voluntarily through institutional and student communication channels. Before accessing the survey items, all participants were required to review the study information and provide informed consent electronically. Participants who declined consent were automatically excluded from further participation. To ensure confidentiality and anonymity, no personally identifiable information was collected.

Statistical Analysis

The statistical analysis for this study was performed using Microsoft Excel (2024) and JASP (University of Amsterdam, version 0.95.4). Categorical variables were expressed as frequencies and percentages, while continuous variables were described using median (IQR). The Shapiro–Wilk test was used to assess the normality of the data, revealing non-normally distributed data for every parameter examined (p<0.05). Statistical tests, such as the Kruskal–Wallis test, were used to compare questionnaire scores among categorical groups with more than two groups. Statistically significant values were considered for p<0.05. For variables showing statistically significant group differences (p<0.05), post-hoc pairwise comparisons were conducted (Dunn’s test) and Rank Epsilon-squared (ε2) effect sizes were calculated. Multiplicity was controlled at the post-hoc stage: the family of comparisons for the Holm–Bonferroni correction was defined as the set of pairwise contrasts across study years for that specific variable. Although individual Likert items yield discrete ordinal data, averaging multiple items across a common domain produces continuous composite scores, making them appropriate for parametric linear regression modeling. Linear regression was performed to evaluate if student’s positive attitude towards AI predicted the perceived clinical utility of AI. It was selected because the primary dependent variable (perceived clinical utility) was operationalized as a continuous composite score. Prior to multivariable regression modeling, standard ordinary least squares assumptions were formally verified: linearity and homoscedasticity were confirmed using inspection of residuals versus predicted values plots, normality of residuals was confirmed through standardized normal Q-Q plots, independence of observations was verified using the Durbin–Watson test (d=2.007, p=0.979), and absence of severe multicollinearity was established via collinearity diagnostics (all condition indices ≤ 17.53) Linear regression allowed us to quantify the direct predictive strength of students’ positive attitude on perceived clinical utility. Multivariable linear regression adjusting for the covariates: gender, year of study, age group, prior AI usage and academic performance was used to further assess the association between the positive attitude towards AI and the perception regarding clinical usage. The Mann–Whitney U-test was implemented to determine differences between two independent groups, based on the non-normally distributed data available. The internal consistency of the questionnaire was assessed using the Cronbach alpha for each domain analysed in the questionnaire. For “Perception regarding AI of medical students”, composed of three items, and alpha of 0.80 has been obtained, while for “Attitude towards AI for medical students”, made of 18 items, the alpha was calculated to be 0.81. Finally, for the “Practical usage of AI apps”, made of 13 questions, the alpha was 0.88 and for the “Clinical applicability of AI apps”, composed of 10 questions, the alpha was calculated to be 0.94, indicating good or excellent internal consistency, depending on the questionnaire domain.

Ethical Considerations

The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. Participation was entirely voluntary, and informed consent was obtained electronically from all participants before questionnaire completion. Data confidentiality and participant anonymity were maintained throughout the study.

Results

The final cohort of students comprised 374 individuals. In Table 1, the gender split is shown, with more women participating in the study (n=256, 68.45%) than men (n=115, 30.74%) in the final database. The remaining students (n=3, 0.80%) either preferred not to disclose their gender or did not fit into any of the aforementioned categories. Among the total number of responders, 241 were 21 or older (64.43%), while only 133 students were younger than 21 years old (35.56%). The best represented study year was the first, with 91 students who completed the questionnaire being enrolled in the first year (24.33%), whereas the fifth year was the worst represented one, with only 35 students (9.35%) being included in this category. Out of the total 374 responders, 359 of them (95.98%) had previously used AI applications, and 76 students (20.32%) were using AI applications daily. When it comes to living arrangements, most students at Victor Babes University who responded to the survey were living at home with their family (29.41%), or they lived alone in an apartment (29.14%).

Table 1.

Demographic Characteristics of the Cohort

Variable Value
Gender, n (%)
Male 115 (30.74)
Female 256 (68.45)
Other 2 (0.53)
I prefer not to mention 1 (0.27)
Age, n (%)
<21 years 133 (35.56)
≥21 years 241 (64.43)
Study year, n (%)
I 91 (24.33)
II 68 (18.18)
III 51 (13.64)
IV 66 (17.64)
V 35 (9.35)
VI 63 (16.84)
Area of residence, n (%)
Urban 298 (79.67)
Rural 76 (20.32)
Current living arrangements, n (%)
University housing 73 (19.51)
Home with family 110 (29.41)
Shared apartment 82 (21.92)
Alone in the apartment 109 (29.14)
Weighted average, n (%)
9.50-10 67 (17.91)
9-9.49 120 (32.08)
8.50–8.99 119 (31.81)
8-8.49 57 (15.24)
<8 11 (2.94)
Use of AI before, n (%)
Yes 359 (95.98)
No 15 (4.01)
Frequency of AI use, n (%)
Every day 76 (20.32)
Often 158 (42.24)
Sometimes 135 (36.09)
Never 5 (1.33%)
Open in a new tab

Table 2 describes the Likert-type responses and their respective frequency of selection for each level, from 1 to 5. The items analysed were based on the students’ perception regarding AI usage, privacy and security, reliability, clinical teaching importance, and the time and effort AI applications can save. When it comes to the direct perception of students of AI apps, they consider them a useful tool to understand medical concepts, with a median of 4 (3–5), and to summarize scientific papers, making them easier to understand, with the same median as mentioned before. With that in mind, most students remain neutral when asked about the correctness of the medical information that AI can provide, with 175 students (46.8%) being part of this category.

Table 2.

Likert-Type Questions and the Frequency of Each Possible Response

Completely Disagree – Level 1 n (%) Partially Disagree – Level 2 n (%) Neutral – Level 3 n (%) Partially Agree – Level 4 n (%) Complete Agreement – Level 5 n (%) Median (IQR)
Perception regarding AI of medical students
How useful do you consider AI applications to be in understanding medical concepts? 4 (1.1) 13 (3.5) 79 (21.1) 136 (36.4) 142 (38.0) 4 (3–5)
How likely do you think it is for AI applications to provide accurate medical information? 5 (1.3) 59 (15.8) 175 (46.8) 102 (27.3) 33 (8.8) 3 (3–4)
Are AI applications a useful tool for summarizing scientific articles? 4 (1.1) 20 (5.3) 83 (22.2) 148 (39.6) 119 (31.8) 4 (3–5)
Attitude towards AI for medical students
I am concerned about the reliability of the information provided by AI applications: 22 (5.9) 85 (22.7) 147 (39.3) 84 (22.5) 36 (9.6) 3 (2–4)
I am worried that relying too much on AI applications will compromise the development of my critical thinking skills: 62 (16.6) 107 (28.6) 95 (25.4) 66 (17.6) 44 (11.8) 3 (2–4)
I have reservations regarding potential security risks involved in using AI applications: 47 (12.6) 86 (23.0) 104 (27.8) 87 (23.3) 50 (13.4) 3 (2–4)
I am concerned about becoming overly reliant on AI applications: 90 (24.1) 80 (21.4) 101 (27.0) 57 (15.2) 46 (12.3) 3 (2–4)
I fear that using AI applications will cause me to lose originality in my academic assignments and preparation: 84 (22.5) 99 (26.5) 90 (24.1) 63 (16.8) 38 (10.2) 3 (2–4)
I fear that the use of AI applications will be prohibited by the faculty/university: 96 (25.6) 100 (26.7) 93 (24.9) 53 (14.2) 32 (8.6) 2 (1–3)
I am concerned about potential privacy risks associated with using AI applications: 54 (14.4) 80 (21.4) 122 (32.6) 75 (20.1) 43 (11.5) 3 (2–4)
I am eager to use AI applications for research or learning: 5 (1.3) 24 (6.4) 93 (24.9) 116 (31.0) 136 (36.4) 4 (3–5)
I believe that AI applications are an important tool for achieving academic success: 20 (5.3) 48 (12.8) 120 (32.1) 101 (27.0) 85 (22.7) 3 (3–4)
I find AI technologies attractive and pleasant to use: 6 (1.6) 20 (5.3) 102 (27.3) 129 (34.5) 117 (31.3) 4 (3–5)
I am always eager to learn new things regarding AI applications: 27 (7.2) 61 (16.3) 104 (27.8) 98 (26.2) 84 (22.5) 3 (3–4)
I trust the opinions of my friends or colleagues regarding the use of AI applications: 23 (6.1) 73 (19.5) 153 (40.9) 92 (24.6) 33 (8.8) 3 (2–4)
I find the use of AI applications helpful in my studies: 8 (2.1) 20 (5.3) 97 (25.9) 131 (35.0) 118 (31.6) 4 (3–5)
I feel comfortable using AI applications to generate content for completing my faculty assignments: 20 (5.3) 48 (12.8) 121 (32.4) 108 (28.9) 77 (20.6) 3 (3–4)
I am concerned about the potential misuse or incorrect use of AI applications in the context of medical education: 8 (2.1) 36 (9.6) 103 (27.5) 107 (28.6) 120 (32.1) 4 (3–5)
I am concerned about the accuracy of the information generated by AI applications regarding complex medical topics: 9 (2.4) 21 (5.6) 100 (26.7) 106 (28.3) 138 (36.9) 4 (3–5)
I believe that AI applications could be a useful resource in preparing for faculty exams: 10 (2.7) 29 (7.8) 101 (27.0) 118 (31.6) 116 (31.0) 4 (3–5)
I believe it is important for medical faculties to establish clear guidelines regarding the use of AI applications: 14 (3.7) 21 (5.6) 95 (25.4) 91 (24.3) 153 (40.9) 4 (3–5)
Practical usage of AI apps
AI applications can help me save time when searching for information: 5 (1.3) 15 (4.0) 42 (11.2) 88 (23.5) 224 (59.9) 5 (4–5)
In my opinion, AI applications represent a valuable resource of accurate information: 11 (2.9) 52 (13.9) 157 (42.0) 104 (27.8) 50 (13.4) 3 (3–4)
I recommend AI applications to my colleagues to ease their academic workload: 19 (5.1) 46 (12.3) 121 (32.4) 86 (23.0) 102 (27.3) 4 (3–5)
AI applications are more useful compared to other information sources I have used in the past: 12 (3.2) 41 (11.0) 114 (30.5) 114 (30.5) 93 (24.9) 4 (3–4)
I have used tools or techniques similar to AI applications in the past: 38 (10.2) 74 (19.8) 114 (30.5) 83 (22.2) 65 (17.4) 3 (2–4)
When I need information to complete academic tasks, I primarily use AI applications: 22 (5.9) 56 (15.0) 122 (32.6) 85 (22.7) 89 (23.8) 3 (3–4)
I frequently use AI applications as an information source for my academic assignments: 20 (5.3) 49 (13.1) 118 (31.6) 106 (28.3) 81 (21.7) 3.5 (3–4)
I believe that relying exclusively on AI applications may affect my critical thinking skills: 18 (4.8) 51 (13.6) 92 (24.6) 97 (25.9) 116 (31.0) 4 (3–5)
I appreciate the accuracy and reliability offered by AI applications: 19 (5.1) 62 (16.6) 177 (47.3) 83 (22.2) 33 (8.8) 3 (3–4)
I believe that using AI applications can save me time and effort in my academic tasks: 7 (1.9) 13 (3.5) 89 (23.8) 100 (26.7) 165 (44.1) 4 (3–5)
It does not take long to learn how to use AI applications: 2 (0.5) 11 (2.9) 58 (15.5) 91 (24.3) 212 (56.7) 5 (4–5)
AI applications do not require advanced technical or information skills to be used: 11 (2.9) 30 (8.0) 75 (20.1) 97 (25.9) 161 (43.0) 4 (3–5)
I am interested in using more sophisticated AI applications to improve my academic results: 8 (2.1) 30 (8.0) 84 (22.5) 94 (25.1) 158 (42.2) 4 (3–5)
Clinical applicability of AI apps
AI applications help me better analyze and understand complex clinical cases: 16 (4.3) 34 (9.1) 102 (27.3) 114 (30.5) 108 (28.9) 4 (3–5)
The use of AI applications contributes to the development of my clinical reasoning skills: 27 (7.2) 62 (16.6) 123 (32.9) 91 (24.3) 71 (19.0) 3 (3–4)
AI applications help me formulate differential diagnoses more efficiently: 13 (3.5) 44 (11.8) 137 (36.6) 96 (25.7) 84 (22.5) 3 (3–4)
I believe that AI applications can improve students’ clinical decision-making capacity: 29 (7.8) 68 (18.2) 124 (33.2) 86 (23.0) 67 (17.9) 3 (2–4)
AI applications facilitate the understanding of medical clinical protocols and guidelines: 8 (2.1) 47 (12.6) 115 (30.7) 102 (27.3) 102 (27.3) 4 (3–5)
Using AI applications increases my confidence in approaching practical clinical scenarios: 22 (5.9) 72 (19.3) 116 (31.0) 103 (27.5) 61 (16.3) 3 (2.25–4)
AI applications can contribute to improving the interpretation of paraclinical investigations (labs, imaging, ECG, etc.): 25 (6.1) 49 (13.3) 125 (33.4) 99 (26.5) 76 (20.3) 3 (3–4)
I believe that AI applications can support the development of clinical competencies necessary for future physicians: 19 (5.1) 57 (15.2) 128 (34.2) 93 (24.9) 77 (20.6) 3 (3–4)
AI applications can facilitate the integration of theoretical knowledge into clinical medical practice: 17 (4.5) 41 (11.0) 118 (31.6) 113 (30.2) 85 (22.7) 4 (3–4)
The integration of AI applications into clinical medical training can improve the quality of medical education: 10 (2.7) 28 (7.5) 120 (32.1) 110 (29.4) 106 (28.3) 4 (3–5)
Open in a new tab

However, on the analysis of students’ attitudes toward AI applications, most of them are not afraid that AI apps will be prohibited by their university or faculty. Students do not seem afraid of their academic life becoming over reliant on AI apps, with a median of 3 (2–4), or the fact that their critical thinking might be affected by relying on AI too much, with a median of 3 (2–4). At the same time, students assessed the potential privacy and security risks as not an issue, but at the same time, were worried about the misuse of AI, while being comfortable with generating content with AI apps. The highest median scores (median 5, 4–5) were observed for items assessing time saved when searching for information and ease of learning to use AI tools. With that in mind, the lowest median in the practical subpart of the questionnaire was a 3, in the subjects’ responses about the quality and the accuracy of information that AI provides. The clinical subpart of Table 2 demonstrated a more reserved pattern of responses compared to the sections discussed before. Items addressing AI-assisted comprehension of complex clinical cases, clinical protocols, and the integration of theoretical knowledge into clinical practice showed relatively higher central tendencies, with median values of 4 (3–5). On the other hand, the items related to more complicated clinical aspects, such as clinical reasoning and decision-making, were indicative of a more neutral perception of the students, with a median of 3.

In Figure 2, the relationship between the positive attitude towards AI and its clinical utility is displayed, as perceived by students. These two composite scores were calculated for each student as the unweighted mean score (the total section score divided by the number of constituent items). Reverse coding was not used, since for the positive attitude towards AI score, only non-negatively worded questions were used in the calculation. The two composite scores included only the positively worded questions, so for the positive attitude, a total of 14 items have been included, across all non-clinical questions, and for the clinical capacity of AI teaching, all 10 items were included in the calculation. Because only Likert type questions were used to calculate this composite scores, the maximum for each student was 5 and the minimum was 1. Using linear regression, the positive attitude towards AI was a significant predictor for students’ perception regarding the clinical utility of AI, (F(1,372)=511.7, p<0.001), and a R2 value of 0.579 and adjusted R2 value of 0.578. For each one-point increase in the positive attitude score, the perceived clinical utility score increased by 0.895 points, 95% CI: 0.817–0.973, B=0.895, β=0.761, SE=0.040, t=22.62. To further explore whether the positive attitude towards AI remains a significantly associated with the perception regarding clinical utility, multivariable linear regression was performed adjusting for gender, age category, study year, prior AI experience, and academic performance. The multivariable model was statistically significant (F(13, 357) = 37.39, p<0.001, R2 = 0.577, adjusted R2 = 0.561), with positive attitude remaining significantly associated with perceived clinical utility (B=0.899, SE =0.042, β= 0.755, t=21.27, p< 0.001). None of the adjusted covariates reached statistical significance (p>0.05).

Figure 2.

Scatter plot: AI attitude (1-5) vs teaching utility (0-5). A scatter plot with a fitted line and a shaded band, plus marginal density shapes along the top and right. The horizontal axis label reads, Positive attitude towards AI, with tick labels 1, 2, 3, 4, 5. The vertical axis label reads, Clinical utility teaching capacity of AI, with tick labels 0, 1, 2, 3, 4, 5. Many circular points are plotted, concentrated most densely from x equals 3 to 5 and y equals 2.5 to 5. A fitted line rises from near x equals 1 and y equals 1 to near x equals 5 and y equals 4.7, with a shaded band around the line. A filled density shape above the plot spans the x range from 1 to 5, with the highest area around x near 4 to 5. A filled density shape to the right of the plot spans the y range from 0 to 5, with the highest area around y near 3 to 5.

Open in a new tab

Relationship between the clinical utility of AI teaching and positive attitude towards it. The circular points represent individual student composite scores. The solid blue line indicates the linear regression fit, and the shaded gray area represents the 95% confidence interval.

In Figure 3, the statistically significant differences across study years were identified for five questionnaire items evaluated using the Kruskal–Wallis tests, with small-to-moderate effect sizes (ε2=0.031–0.052). Prior utilization of similar AI tools differed significantly across cohorts (H(5)=19.40, p=0.002, ε2=0.052, 95% CI: 0.028–0.118), with post-hoc Dunn’s tests (Holm–Bonferroni adjusted) indicating that 1st-year students reported significantly higher prior tool use than 2nd-year (padj=0.024), 4th-year (padj=0.012), and 6th-year students (padj=0.009). Evaluation of AI as a valuable source of information also varied significantly across years (H(5)=17.70, p=0.003, ε2=0.047, 95% CI: 0.020–0.110), driven by 1st-year students assigning significantly higher value than 4th-year students (padj=0.009). Regarding workflow efficiency, significant differences emerged for perceived time savings when searching for information (H(5)=14.26, p=0.014, ε2=0.038, 95% CI: 0.021–0.095) and perceived effort reduction in academic tasks (H(5)=13.18, p=0.022, ε2=0.035, 95% CI: 0.016–0.094); 6th-year students perceived significantly greater time savings than 3rd-year students (padj=0.007), as well as greater effort savings compared to both 3rd-year (padj=0.040) and 4th-year students (padj=0.022). Finally, reliance on peer opinions regarding AI differed significantly across cohorts (H(5)=11.38, p=0.044, ε2=0.031, 95% CI: 0.012–0.090), with 1st-year students placing significantly higher trust in peer opinions than 5th-year students (padj=0.043). All other questionnaire items yielded non-significant omnibus Kruskal–Wallis results across study years (p>0.05), and no other pairwise contrasts reached statistical significance following multiplicity correction.

Figure 3.

A set of five bar charts showing artificial intelligence questionnaire ratings by study year. The text describes five bar charts showing various perceptions of AI by study year. Image A: Trust in peers′ opinions on AI, with scores from 1st to 6th year being 3.4, 3.2, 3.0, 3.1, 2.8 and 3.1 respectively. Image B: Usage of similar tools in the past, with scores of 3.6, 3.0, 3.4, 3.0, 3.1 and 2.9. Image C: AI′s role in saving effort in academic tasks, with scores of 4.1, 4.1, 3.8, 3.9, 4.0 and 4.4. Image D: AI′s efficiency in saving time when searching for information, with scores of 4.4, 4.3, 4.0, 4.3, 4.4 and 4.6. Image E: AI as a valuable information source, with scores of 3.6, 3.5, 3.3, 3.0, 3.1 and 3.5. Each chart includes error bars.

Open in a new tab

Kruskal–Wallis analysis based on study year for the most relevant differences between groups.

Figure 4 illustrates the highest agreement rates among students regarding Likert-type items. Agreement has been defined as a response of 4 or 5 on the Likert scale. The most agreed on item was the capacity of AI to save time when searching for information, with 83.42% of students agreeing it is useful. Secondly, the majority of students agreed that AI apps do not take long to learn how to use. Out of the total, 74.33% of students agreed that AI apps are useful in understanding medical concepts, while 71.39% of responders considered AI useful in summarizing scientific articles.

Figure 4.

A horizontal bar graph showing student agreement rates for statements about artificial intelligence applications. A horizontal bar graph displays eight statements about AI applications with agreement percentages. The x-axis is labeled percent, ranging from 0 to 100 in 10 percent steps. The y-axis lists Likert-type statements. Bar values from top to bottom: AI apps can save time when searching for information, 83.42%. They are quick to learn, 81.02%. Useful for understanding medical concepts, 74.33%. Helpful for summarizing scientific articles, 71.39%. Believed to save time and effort in academic tasks, 70.86%. Do not require advanced technical skills, 68.98%. Users are eager to learn more about AI, 67.38%. Interested in using sophisticated AI for better academic results, 67.38%.

Open in a new tab

The highest agreement rates among students based on the Likert-type questions (Agreement means a score of 4 – “Partially agree” or 5 – “Totally agree” on the Likert scale).

Discussion

The present study included a heterogeneous cohort of Romanian medical students from all six years of training, predominantly female and originating from urban areas. Similar demographic distributions have been reported in recent studies evaluating medical students’ perceptions of artificial intelligence (AI) and digital educational technologies, where female participants represented most respondents.24,25 Most participants demonstrated high academic performance, with weighted averages above 8.50. However, the present study did not directly assess academic motivation or engagement. The inclusion of both preclinical and clinical students allowed attitudes toward AI to be examined across different stages of medical education, consistent with previous cross-sectional studies investigating AI adoption among medical students.24,26

One of the most important findings was the high prevalence of prior AI use among Romanian medical students, with nearly all participants reporting previous experience with AI-based applications and most indicating daily or frequent utilization. The high prevalence of AI use observed in this cohort is consistent with the increasing adoption of generative AI technologies reported among healthcare students and young professionals.4,21,22

Beyond this high level of AI adoption, the regression analysis identified a strong positive association between students’ attitudes toward AI and their perception of its clinical utility. In the regression model, positive attitude toward AI was specified as the independent variable, whereas perceived clinical utility was the dependent variable. Students with more favorable attitudes toward AI tended to report greater perceived clinical utility, with the univariable model explaining approximately 57.9% of the variance in perceived clinical utility. This association remained statistically significant after adjustment for gender, age category, study year, prior AI experience, and academic performance.

These findings are consistent with previous studies showing that attitudes toward AI, perceived usefulness, performance expectancy, and practical applicability are closely related to the acceptance and perceived value of AI technologies among medical students and healthcare professionals.4,13,21–24,27,28 Previous research has also described AI applications as potentially valuable complementary educational resources, particularly for supporting access to information, independent learning, and clinically oriented educational activities.5,10,25,29

Previous research suggests that healthcare trainees may report more favorable attitudes toward AI when these technologies are perceived as supportive rather than substitutive tools in clinical practice.22,24,30 In addition, medical students have expressed interest in the integration of structured AI-related educational content into undergraduate curricula, particularly in areas involving clinical reasoning, evidence-based decision-making, digital literacy, and healthcare innovation.20,23,25,31–34

Importantly, because the present study used a cross-sectional design and both constructs were assessed simultaneously, the observed association should not be interpreted as evidence of causality or temporal direction. Although the statistical model specified positive attitude toward AI as the independent variable and perceived clinical utility as the dependent variable, a reciprocal relationship cannot be excluded. Perceived clinical usefulness may also contribute to favorable attitudes toward AI; however, this direction was not evaluated as the primary regression model in the present study. Longitudinal studies are therefore needed to clarify the temporal relationship between attitudes toward AI and perceptions of its clinical usefulness. Moreover, because both constructs were measured simultaneously using the same self-reported questionnaire, common-method variance and conceptual overlap between the two composite scores may have contributed to the strength of the observed association.

Differences between study years were limited to specific questionnaire items and should not be interpreted as a general trend across academic years. Previous studies have suggested that students with greater clinical exposure may perceive AI technologies as more applicable to real-world medical practice and clinical decision-making;13,22,25,28,35,36 however, the present findings do not support extrapolating such a pattern across all study years or questionnaire domains. Therefore, the observed differences should be interpreted strictly at the item level.

The lower score observed among third-year students for the specific time- and effort-saving item may reflect differences in educational stage or clinical exposure; however, these factors were not directly assessed, and no causal explanation can be established from the present data.

Another important observation of the present study was the generally favorable perception regarding the educational and practical utility of AI applications among Romanian medical students. Most participants considered AI tools useful for understanding medical concepts, summarizing scientific literature, supporting academic activities, and facilitating access to medical information.4,10 Particularly high agreement rates were identified for statements related to time efficiency, ease of use, and the ability of AI applications to simplify information retrieval processes. These findings indicate that students perceive AI technologies as practical educational support tools that may facilitate learning activities and academic productivity; however, these perceptions should not be interpreted as evidence of objectively improved educational outcomes. Similar perceptions have been reported in previous international studies, where medical students and educators frequently highlighted accessibility, educational efficiency, and the potential of AI-assisted technologies to support learning and future medical training.21,24,37

Although students generally demonstrated favorable attitudes toward AI, responses to potential risks were mixed. Fewer than one third agreed that excessive reliance on AI could compromise their critical thinking skills or that they were concerned about becoming overly reliant on AI applications, while substantial proportions were neutral or disagreed with these statements. Concerns regarding reliability, privacy, academic integrity, and misuse nevertheless remained relevant across individual questionnaire items.16–20,38

Another relevant finding was the high proportion of students supporting the implementation of institutional guidelines regarding AI use in medical education. This finding is consistent with increasing attention to the ethical and professional implications of AI-assisted learning and clinical decision-making.20,31 Moreover, students appeared particularly interested in using more advanced AI applications to support their academic activities and clinical learning, suggesting substantial openness toward future curricular integration of AI-related educational content.32,33 These findings are consistent with recent studies indicating that medical students increasingly favor structured AI literacy training and formal educational frameworks addressing the responsible use of AI technologies in healthcare education.34,39

Regarding the clinical dimension, most participants perceived AI applications as potentially useful for understanding complex clinical cases, supporting differential diagnosis formulation, and facilitating the integration of theoretical knowledge into clinically oriented learning activities. These findings reflect students’ perceptions of potential clinical educational utility and should not be interpreted as evidence that AI objectively improves clinical skills, diagnostic performance, or the quality of clinical training. However, responses regarding AI’s direct contribution to clinical reasoning and decision-making abilities were comparatively more moderate, suggesting that students may still perceive AI primarily as a complementary support tool rather than an autonomous clinical resource.22,25,27,35 This pattern suggests that favorable perceptions of AI coexist with recognition of the importance of human judgment, clinical expertise, and critical reasoning in medical practice.36 Similar attitudes have been described in recent studies emphasizing that medical students generally support AI integration into clinical education while simultaneously recognizing the irreplaceable role of physicians in complex clinical decision-making processes.

Finally, the widespread adoption of AI applications in this cohort coexisted with a generally low perceived technical barrier to their use. Together with persistent concerns regarding reliability and ethical implications, these findings support the relevance of structured educational programs addressing AI literacy, critical appraisal, and responsible AI use among future healthcare professionals.

Study Limitations

This study has several limitations that should be acknowledged. First, the cross-sectional design does not allow the establishment of causal relationships between students’ characteristics, AI use, and their perceptions regarding the educational and clinical utility of AI applications. Second, the study was conducted in a single medical university in Romania; therefore, the findings may not be fully generalizable to all Romanian medical students or to students from other educational and cultural contexts. Third, the data was collected using a self-administered questionnaire, which may be subject to response bias, social desirability bias, and individual differences in the interpretation of Likert-scale items. Also, while internal consistency was demonstrated across all four questionnaire domains using Cronbach’s alpha, formal construct validation through exploratory or confirmatory factor analysis was not undertaken for this modified instrument. Because of the online method used to distribute the questionnaire self-selection and selection biases have to be acknowledged as limitations.

Another limitation is that the study evaluated students’ perceptions and self-reported use of AI applications rather than objective measures of academic performance, clinical reasoning, or practical skill development. Additionally, because positive attitudes and perceived clinical utility were evaluated simultaneously using the same Likert-scale questionnaire, the potential for common-method variance and conceptual overlap between these domains should be considered when interpreting the strong association (R2=0.579). Minor numerical discrepancies between prior AI exposure (n = 15 non-users) and reported usage frequency (n = 5, “Never”) reflect raw participant response inconsistencies inherent to un-branched online survey administration. Therefore, although students reported favorable attitudes toward AI tools, the present findings cannot confirm whether AI use directly improves educational or clinical outcomes. In addition, the questionnaire assessed AI applications in general and did not differentiate between specific tools, such as ChatGPT, Gemini, Copilot, or other AI-based platforms, which may differ in accuracy, accessibility, functionality, and educational value. Finally, because AI technologies are rapidly evolving, students’ perceptions may change over time as these tools become more advanced, more widely integrated into medical curricula, and subject to clearer institutional regulations.

Despite these limitations, the present study provides valuable preliminary evidence regarding Romanian medical students’ perceptions of AI applications in medical education and highlights important directions for future multicenter and longitudinal research.

Future Directions

Future research should further investigate the long-term educational and clinical impact of artificial intelligence applications in medical training, particularly regarding their influence on clinical reasoning, practical competencies, decision-making abilities, and the development of non-technical skills among future physicians. Although interest in AI-assisted learning has expanded considerably worldwide, important regional differences may influence the adoption, perception, and effective integration of these technologies into medical education.

In Eastern European countries, including Romania, the implementation of AI-based educational tools may be influenced by socio-cultural, economic, technological, and institutional factors such as variability in digital infrastructure, unequal access to educational resources, differences in curricular modernization, and limited formal training in digital literacy and artificial intelligence.

Therefore, future multicenter and longitudinal studies involving larger and more diverse student populations from Romania and other Eastern European countries are necessary to better understand regional attitudes, educational needs, and potential barriers related to AI integration in medical education. Additional research should also evaluate how cultural perceptions, educational traditions, and institutional policies may influence students’ trust in AI technologies and their willingness to incorporate these tools into clinical learning and professional development. Furthermore, the development of evidence-based educational strategies, ethical guidelines, and structured AI literacy programs for both students and educators will likely represent an essential step toward the responsible and effective implementation of artificial intelligence within modern medical curricula. Such initiatives may contribute not only to improving educational quality and adaptability but also to preparing future healthcare professionals for the increasing digitalization of clinical practice and healthcare systems worldwide.

Conclusion

Medical students from a single Romanian medical university showed high familiarity with AI applications and generally favorable perceptions of their educational and clinical utility. Positive attitudes toward AI were strongly associated with greater perceived clinical utility, including after adjustment for demographic and academic covariates; however, given the cross-sectional design, this association should not be interpreted as causal or temporal.

Perceptions differ across specific questionnaire items and study years, and longitudinal research is needed to determine whether attitudes change over time. Concerns regarding reliability, accuracy, ethics, and misuse were also identified. These findings suggest that structured AI literacy training and appropriate institutional guidance may support the responsible use of AI in undergraduate medical education.

Funding Statement

The authors gratefully acknowledge the financial support provided by Victor Babeș University of Medicine and Pharmacy Timișoara, which covered the publication costs associated with this study.

Data Sharing Statement

The data supporting the findings of this study are not publicly accessible due to institutional restrictions and the need to protect participant confidentiality and privacy. However, anonymized data may be made available by the corresponding author upon reasonable request and subject to approval in accordance with institutional ethical guidelines and applicable data protection legislation. All data requests can be made to Bogdan Chiu.

Ethics Approval

This study was conducted in accordance with the ethical standards outlined in the Declaration of Helsinki. Ethical approval was granted by the Ethics Committee of Victor Babeș University of Medicine and Pharmacy Timișoara (approval no. 90/01.10.2025 rev 15.05.2026). All participants received detailed information regarding the objectives and procedures of the study, and participation was entirely voluntary. Written informed consent was obtained from all participants before inclusion in the study. To ensure confidentiality and data protection, all collected data were anonymized and processed in accordance with applicable ethical and institutional regulations.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that there are no conflicts of interest.

References

  • 1.Gordon M, Daniel M, Patricio M, et al. A scoping review of artificial intelligence in medical education BEME Guide No 84. Med Teach. 2024;46(6):614–17. [DOI] [PubMed] [Google Scholar]
  • 2.Choi-Lundberg DL, Cuellar WA, Williams AM. Online and technology-enhanced learning in medical education. Clin ahmedTeach. 2023;20(5):e13620 [Google Scholar]
  • 3.Lattouf OM. Impact of digital transformation on the future of medical education and practice. J Card Surg. 2022;37(8):2239–2247. doi: 10.1111/jocs.16642 [DOI] [PubMed] [Google Scholar]
  • 4.Sallam M. ChatGPT utility in healthcare education, research, and practice: systematic review on the promising perspectives and valid concerns. Healthcare. 2023;11(6):887. doi: 10.3390/healthcare11060887 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Kung TH, Cheatham M, Medenilla A, et al. Performance of ChatGPT on USMLE: potential for AI-assisted medical education using large language models. PLOS Digit Health. 2023;2(2):e0000198. doi: 10.1371/journal.pdig.0000198 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Gilson A, Safranek CW, Huang T, et al. How does ChatGPT perform on the United States Medical Licensing Examination? The implications of large language models for medical education and knowledge assessment. JMIR Med Educ. 2023;9:e45312. doi: 10.2196/45312 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Khan RA, Jawaid M, Khan AR, Sajjad M. ChatGPT - Reshaping medical education and clinical management. Pak J Med Sci. 2023;39(2):605–607. doi: 10.12669/pjms.39.2.7653 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Lee H. The rise of ChatGPT: exploring its potential in medical education. Anat Sci Educ. 2023;16(5):899–901. [DOI] [PubMed] [Google Scholar]
  • 9.O’Connor S. Open artificial intelligence platforms in nursing education: tools for academic progress or abuse? Nurse Educ Pract. 2023;66:103537. doi: 10.1016/j.nepr.2022.103537 [DOI] [PubMed] [Google Scholar]
  • 10.Boscardin CK, Gin B, Golde PB, Hauer KE. ChatGPT and generative artificial intelligence for medical education: potential impact and opportunity. Acad Med. 2024;99(1):22–27. doi: 10.1097/ACM.0000000000005439 [DOI] [PubMed] [Google Scholar]
  • 11.Wong K, Fayngersh A, Traba C, Cennimo D, Kothari N, Chen S. Using ChatGPT in the development of clinical reasoning cases: a qualitative study. Cureus. 2024;16(5):e61438. doi: 10.7759/cureus.61438 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Skryd A, Lawrence K. ChatGPT as a tool for medical education and clinical decision-making on the wards: case study. JMIR Form Res. 2024;8:e51346. doi: 10.2196/51346 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Wu Y, Zheng Y, Feng B, Yang Y, Kang K, Zhao A. Embracing ChatGPT for medical education: exploring its impact on doctors and medical students. JMIR Med Educ. 2024;10:e52483. doi: 10.2196/52483 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Misra SM, Suresh S. Artificial intelligence and objective structured clinical examinations: using ChatGPT to revolutionize clinical skills assessment in medical education. J Med Educ Curric Dev. 2024;11:23821205241263475. doi: 10.1177/23821205241263475 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Pham TD, Karunaratne NS, Exintaris B, et al. The impact of generative AI on health professional education: a systematic review in the context of student learning. Med Educ. 2025;59(12):1280–1289. doi: 10.1111/medu.15746 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Alkaissi H, McFarlane SI. Artificial hallucinations in ChatGPT: implications in scientific writing. Cureus. 2023;15(2):e35179. doi: 10.7759/cureus.35179 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Cascella M, Montomoli J, Bellini V, Bignami E. Evaluating the feasibility of ChatGPT in healthcare: an analysis of multiple clinical and research scenarios. J Med Syst. 2023;47(1):33. doi: 10.1007/s10916-023-01925-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Mijwil MM, Aljanabi M, Ali AH. ChatGPT and medical education: a review of applications, limitations, and ethical concerns. Inform Med Unlocked. 2023;41:101324. [Google Scholar]
  • 19.De Angelis L, Baglivo F, Arzilli G, et al. ChatGPT and the rise of large language models: the new AI-driven infodemic threat in public health. Front Public Health. 2023;11:1166120. doi: 10.3389/fpubh.2023.1166120 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Masters K. Ethical use of artificial intelligence in health professions education AMEE Guide No 158. Med Teach. 2024;46(2):123–134. [DOI] [PubMed] [Google Scholar]
  • 21.Fijačko N, Gosak L, Cilar L, Novšak A, Mihelič Zajec N, Stiglic G. Exploring the perceived usefulness of ChatGPT in healthcare education among students: a cross-sectional study. Nurs Educ Today. 2024;134:106051. [Google Scholar]
  • 22.D PDS, Giese D, Brodehl S, et al. Medical students’ attitude towards artificial intelligence: a multicentre survey. Eur Radiol. 2020;30(3):1640–1646. [DOI] [PubMed] [Google Scholar]
  • 23.AlAlwan AQM, Lone A. ChatGPT in medical education: a cross-sectional analysis of usage, attitudes, perceptions, and practices among Saudi medical students. Front Artif Intell. 2026;9:1751367. PMID: 42099706; PMCID: PMC13144081. doi: 10.3389/frai.2026.1751367 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Buabbas AJ, Al-Shawaf HM, Almajran AA, et al. Investigating students’ perceptions towards artificial intelligence in medical education. Front Public Health. 2023;11:1150850. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Robleto E, Thompson A, Solano D, et al. medical students’ perceptions of an artificial intelligence (AI)-assisted diagnostic tool for clinical decision-making. BMC Med Educ. 2024;24(1):149. doi: 10.1186/s12909-024-05089-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Chakri I, Bakhouch K, Elhangouche A, et al. Knowledge and perceptions of artificial intelligence among medical students in Morocco: a cross-sectional study. Cureus. 2025;17(2):e79844. doi: 10.7759/cureus.79844 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Sit C, Srinivasan R, Amlani A, et al. Attitudes and perceptions of UK medical students towards artificial intelligence and radiology: a multicentre survey. Insights Imaging. 2020;11(1):14. doi: 10.1186/s13244-019-0830-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Jebreen K, Radwan E, Kammoun-Rebai W, et al. Perceptions of undergraduate medical students on artificial intelligence in medicine: mixed-methods survey study from Palestine. BMC Med Educ. 2024;24(1):507. doi: 10.1186/s12909-024-05465-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Oh S, Kim JH, Choi SW, Lee HJ, Hong J, Kwon SH. Physician confidence in artificial intelligence: an online mobile survey. J Med Internet Res. 2019;21(3):e12422. doi: 10.2196/12422 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Yeo YH, Samaan JS, Ng WH, et al. Assessing the performance of ChatGPT in answering questions regarding cirrhosis and hepatocellular carcinoma. Clin Mol Hepatol. 2023;29(3):721–732. doi: 10.3350/cmh.2023.0089 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Mehta N, Harish V, Bilimoria K, Morgado F, Ginsburg S, Law M. Knowledge and attitudes on artificial intelligence in healthcare: a provincial survey study of medical students. Med Educ Online. 2021;26(1):1914303. [Google Scholar]
  • 32.Karaca O, Çalışkan SA, Demir K. Medical artificial intelligence readiness scale for medical students (MAIRS-MS) – development, validity and reliability study. BMC Med Educ. 2021;21(1):112. doi: 10.1186/s12909-021-02546-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Tolentino R, Baradaran A, Gore G, Pluye P, Abbasgholizadeh-Rahimi S. Curriculum frameworks and educational programs in artificial intelligence in medical education: a scoping review. JMIR Med Educ. 2024;10:e52271. doi: 10.2196/54793 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Pupic N, Fan KS, Tassone M, et al. An evidence-based approach to artificial intelligence in undergraduate medical education: systematic review. Med Teach. 2023;45(11):1177–1184. doi: 10.1080/0142159X.2023.2198665 [DOI] [PubMed] [Google Scholar]
  • 35.Weidener L, Fischer M. Artificial intelligence in medicine: cross-sectional study among medical students in Germany, Austria, and Switzerland on AI in medicine and AI ethics education. JMIR Med Educ. 2024;10:e54724. doi: 10.2196/51247 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Alkhaaldi SMI, Kaloti R, AlMuhaideb S, Aljamaan F, Alabdulkarim Y. Medical student experiences and perceptions of ChatGPT and artificial intelligence: cross-sectional study. Med Educ Online. 2023;28(1):2287911. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Mir MM, Mir GM, Raina NT, et al. Application of artificial intelligence in medical education: current scenario and future perspectives. J Adv Med Educ Prof. 2023;11(3):133–140. doi: 10.30476/JAMP.2023.98655.1803 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Dave T, Athaluri SA, Singh S. ChatGPT in medicine: an overview of its applications, limitations, and ethical considerations. Front Artif Intell. 2023;6:1169595. doi: 10.3389/frai.2023.1169595 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Chan KS, Zary N. Applications and challenges of implementing artificial intelligence in medical education: integrative review. JMIR Med Educ. 2019;5(1):e13930. doi: 10.2196/13930 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The data supporting the findings of this study are not publicly accessible due to institutional restrictions and the need to protect participant confidentiality and privacy. However, anonymized data may be made available by the corresponding author upon reasonable request and subject to approval in accordance with institutional ethical guidelines and applicable data protection legislation. All data requests can be made to Bogdan Chiu.

Articles from Advances in Medical Education and Practice are provided here courtesy of Dove Press

RESOURCES