Adoption of generative AI chatbots among medical postgraduates at two universities in China: patterns, attitudes, and concerns - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice BMC Med Educ . 2026 Mar 18;26:626. doi: 10.1186/s12909-026-08947-9 Search in PMC Search in PubMed View in NLM Catalog Add to search Adoption of generative AI chatbots among medical postgraduates at two universities in China: patterns, attitudes, and concerns Lijun Zhao Lijun Zhao 1 Department of General Practice, General Practice Medical Center, West China Hospital of Sichuan University, No. 37, Guoxue Alley, Chengdu, Sichuan Province 610041 China Find articles by Lijun Zhao 1 , Qianqian Han Qianqian Han 2 Department of Pathology, West China Hospital of Sichuan University, Chengdu, 610041 China Find articles by Qianqian Han 2 , Peijuan Li Peijuan Li 1 Department of General Practice, General Practice Medical Center, West China Hospital of Sichuan University, No. 37, Guoxue Alley, Chengdu, Sichuan Province 610041 China Find articles by Peijuan Li 1 , Hua Dai Hua Dai 1 Department of General Practice, General Practice Medical Center, West China Hospital of Sichuan University, No. 37, Guoxue Alley, Chengdu, Sichuan Province 610041 China Find articles by Hua Dai 1, ✉ , Xuegui Ju Xuegui Ju 3 The First Affiliated Hospital of Chengdu Medical College, No. 278 Middle of Baoguang Road, Xindu District, Chengdu, Sichuan 610500 China Find articles by Xuegui Ju 3, ✉ Author information Article notes Copyright and License information 1 Department of General Practice, General Practice Medical Center, West China Hospital of Sichuan University, No. 37, Guoxue Alley, Chengdu, Sichuan Province 610041 China 2 Department of Pathology, West China Hospital of Sichuan University, Chengdu, 610041 China 3 The First Affiliated Hospital of Chengdu Medical College, No. 278 Middle of Baoguang Road, Xindu District, Chengdu, Sichuan 610500 China ✉ Corresponding author. Received 2025 Aug 26; Accepted 2026 Feb 27; Collection date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ . PMC Copyright notice PMCID: PMC13085431 PMID: 41851693 Abstract Background Generative artificial intelligence (AI) chatbots are gaining attention in medical education globally for their potential to support academic writing, clinical reasoning, and personalized learning. However, little is known about their adoption, benefits, and concerns among Chinese postgraduate medical students, particularly across distinct clinical- and academic-track programs that characterize China’s unique medical education system. Methods A cross-sectional survey was conducted among 340 postgraduate medical students from two universities in Chengdu, China. A structured questionnaire assessed AI awareness, usage patterns, perceived benefits, attitudes, and concerns. Descriptive statistics, subgroup analyses, and Pearson correlation analyses were applied to examine differences by gender and degree type. Results Most students (82.9%) reported strong AI awareness, with DeepSeek (90.9%) and ChatGPT (55.2%) most frequently used. Common applications included literature review (61.5%), exam preparation (55.0%), and clinical case analysis (48.5%). Reported benefits encompassed faster information retrieval (70.8%) and improved writing precision (64.3%). Overall satisfaction was high (mean 4.4/5), and 84.0% supported curriculum integration. Female students use AI more frequently but expressed greater concerns, clinical-track students demonstrated higher awareness than academic-track students. Awareness, usage, and attitudes showed positive correlations, while concerns about accuracy and ethics remained independent of prior exposure. Conclusions This study at two universities in Chengdu demonstrates widespread adoption and favorable perceptions of generative AI chatbots among postgraduate medical students. Findings suggest that structured curricular integration, accompanied by ethical safeguards addressing China-specific challenges, may maximize benefits while mitigating risks. Multi-center studies are needed to validate these regional patterns nationally. Supplementary Information The online version contains supplementary material available at 10.1186/s12909-026-08947-9. Keywords: Generative artificial intelligence, Chatbots, Medical education, Perceptions, Postgraduate medical students, China Introduction Artificial Intelligence (AI) refers to computational systems capable of performing tasks that typically require human intelligence, encompassing domains such as machine learning (ML), natural language processing (NLP), and deep learning. In medicine, AI has progressed from theoretical models to clinically integrated tools, transforming fields such as medical imaging analysis (e.g., tumor detection in radiology), predictive diagnostics (e.g., sepsis risk stratification), drug discovery, and personalized treatment planning [ 1 , 2 ]. The World Health Organization has acknowledged AI’s potential to address global healthcare disparities, particularly in resource-limited settings [ 3 ]. Generative AI—an emerging subset of AI capable of producing novel text, images, or structured data—represents a significant frontier with transformative implications for both clinical practice and medical education. Medical chatbots, powered by NLP, have traditionally been used to simulate human conversation, assist in scheduling, and manage electronic health records. More recently, generative AI chatbots such as ChatGPT and DeepSeek have demonstrated the ability to synthesize complex, context-aware responses to open-ended academic and clinical queries, thereby extending their applications beyond routine administrative tasks [ 4 ]. AI-based educational tools, including intelligent virtual patient simulations, adaptive learning platforms, and generative question–answer systems, are increasingly being integrated into medical curricula, with evidence suggesting improvements in teaching efficiency and student engagement [ 5 ]. The Technology Acceptance Model (TAM) has been extensively applied to understand users’ acceptance of educational technologies including e-learning and other learning tools [ 6 ]. The Diffusion of Innovation theory explains how new ideas and technologies spread over time within social systems through factors such as relative advantage, compatibility, complexity, trialability, and observability [ 6 ]. Together, these frameworks provide a robust theoretical foundation for examining medical students’ adoption patterns, attitudes, and behavioral intentions toward generative AI chatbots. International studies have begun documenting AI adoption patterns among medical students across diverse contexts. A European cross-sectional study across 192 medical faculties in 63 countries reported that 58.3% of students had used AI tools, with higher adoption rates among students in higher income countries and those with prior technology training [ 7 ]. Middle Eastern studies have revealed comparable patterns: Saudi medical students demonstrated 73.2% awareness of ChatGPT with generally positive attitudes toward its educational applications, though significant concerns about accuracy and ethical implications persisted [ 8 ]. The adoption of generative AI chatbots in medical education offers several advantages, including enhanced access to information, real-time language refinement, and personalized feedback [ 9 ]. However, legitimate concerns persist, particularly regarding the generation of inaccurate information (“hallucinations”), the potential for plagiarism, biases in training data, and copyright violations [ 10 , 11 ]. Such issues may influence the perceptions, expectations, and attitudes of both students and educators, potentially shaping the effectiveness of AI-enhanced teaching [ 12 ]. Despite the growing international evidence base [ 7 ], systematic investigations of AI adoption in Chinese medical education remain scarce. China’s medical education system differs substantially from Western models in its clear bifurcation between clinical-track programs (emphasizing hands-on hospital training with direct pathways to residency certification) and academic-track programs (focusing on laboratory research and scientific publication with additional training required for clinical practice). Moreover, China’s unique linguistic context—where students frequently navigate between Chinese-language instruction and English-language research literature—may create distinct patterns of AI tool usage compared to primarily English-speaking educational environments. Understanding how these factors influence AI adoption among Chinese postgraduate medical students is essential for developing evidence-based policies that are both globally informed and locally appropriate. The present cross-sectional study investigates adoption patterns of generative chatbots among Chinese medical students, evaluates their attitudes and concerns, and explores behavioral intentions toward AI in both academic and clinical learning contexts, thereby providing evidence tailored to China’s unique medical education landscape. Materials and methods Study design This study employed convenience sampling to recruit postgraduate medical students from two universities in Chengdu, Sichuan Province, China: West China Hospital of Sichuan University and The First Affiliated Hospital of Chengdu Medical College. These institutions were selected based on accessibility and willingness to participate. While both are major academic medical centers in southwestern China. This cross-sectional survey was aimed to assess their knowledge, attitudes, concerns, and behavioral intentions toward the use of generative AI chatbots in academic and educational contexts. Data were collected between June and August 2025. The questionnaire was distributed to students across different specialties and stages of postgraduate training. Eligible participants were current postgraduate medical students who had access to electronic devices and consented to participate in the study. Exclusion criteria were applied to individuals who were not enrolled in medical universities, lacked internet access, or declined participation. This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board at the West China Hospital of Sichuan University (Approval No. 2023 − 1066). All participants provided informed consent prior to their inclusion in the study. Electronic consent was obtained from every participant in accordance with institutional and ethical guidelines. Electronic consent was obtained from every participant at the beginning of the online survey. Participants encountered the consent form as the first page, which included: study purpose, procedures, voluntary participation statement, right to withdraw, confidentiality assurances, data protection measures, and contact information. Participants were required to affirmatively check a box stating ‘I have read and understood the above information and agree to participate’ before accessing the questionnaire. Questionnaire content The online survey was developed in accordance with the Checklist for Reporting Results of Internet E-Surveys (CHERRIES) [ 13 ] and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [ 14 ]. The final questionnaire comprised 40 items organized into seven sections and was administered via the WJX platform ( https://www.wjx.cn/ ). The survey addressed the following domains: participant demographics, awareness and understanding of AI, patterns of AI chatbot use, educational applications of AI, perceptions of generative AI, and students’ attitudes, concerns, and expectations regarding AI in medical education and academic research. The survey link was disseminated through community engagement groups on the WeChat social media platform (Tencent Inc., Shenzhen, China). Questionnaire development and content validity The questionnaire was developed in several steps to ensure that the items adequately reflected the intended constructs. First, the research team conducted a narrative review of published studies on AI in medical and health-profession education and on instruments measuring technology acceptance, attitudes and concerns toward AI [ 12 , 15 , 16 ]. Based on this literature and our conceptual framework, an initial pool of items was drafted in Chinese to cover five domains: attitudes toward AI integration in medical education, general concerns about AI use, concerns regarding potential risks in medical education, behavioral intentions to use AI tools, and perceived changes in ability after using AI chatbots. Content validity was then evaluated by a panel of three experts: a senior clinician involved in postgraduate training, a medical education specialist, and a researcher with experience in AI-assisted teaching and quantitative survey design. Each expert independently rated the relevance of every item to its target construct on a 4-point scale (1 = not relevant, 4 = highly relevant) and provided qualitative comments on clarity and wording. For each item, an item-level content validity index (I-CVI) was calculated as the proportion of experts assigning a rating of 3 or 4. Across the 23 items, I-CVI values ranged from 0.67 to 1.00. For each multi-item scale, a scale-level content validity index (S-CVI/Ave) was computed by averaging the I-CVI values: 0.95 for attitudes toward AI integration, 1.00 for general concerns, 0.83 for medical education concerns, 1.00 for behavioral intentions, and 0.92 for perceived ability changes, with an overall S-CVI/Ave of 0.94. These values indicate acceptable to excellent content validity. Items with lower I-CVI values (0.67) were discussed and refined for clarity but retained because at least two of three experts judged them to be relevant. A pilot test was subsequently conducted with postgraduate medical students not included in the main sample to assess comprehensibility, layout and completion time. Only minor wording refinements were required, and no structural changes to the domains were made. In the main study sample, internal consistency of the five multi-item scales was good to excellent, with Cronbach’s α values of 0.886 (attitudes toward AI integration), 0.901 (general concerns), 0.914 (medical education concerns), 0.922 (behavioral intentions) and 0.858 (perceived ability changes) (Supplementary Table 1). Data cleaning and quality control To ensure data integrity and reliability, responses underwent a structured cleaning process. Exclusion criteria were applied as follows: (1) Responses with more than one unanswered item were excluded. (2) Submissions demonstrating uniform or highly repetitive answer patterns across Likert-scale items (suggesting inattentive responding) were flagged. (3) To validate an appropriate completion time threshold, we conducted a comprehensive analysis of the completion time distribution in the full dataset ( n = 345). The median completion time was 116 s (IQR: 83–159 s), with a mean of 138 s (SD = 115). Using Tukey’s method (Q1–1.5 × IQR), the statistical lower bound for outlier detection was − 31 s, indicating that all completion times in our sample fell within the statistically plausible range. An initial pilot test with five team members suggested a 50-second minimum completion time. While this pilot-based estimate had limitations (small, non-independent sample), the distribution analysis confirmed that 50 s represented an extremely rapid completion pace—well below the 25th percentile (83 s). Notably, only 5 responses (1.4%) met multiple quality concerns simultaneously: completion time under 50 s combined with either uniform response patterns or substantial missing data. These convergent indicators supported their exclusion. Sensitivity analysis of exclusion criteria: We compared key findings between the full dataset ( n = 345) and the analytical sample ( n = 340) to assess potential bias introduced by exclusions. Results showed negligible differences in demographic distributions, mean scores, and correlation coefficients (all differences < 0.05), confirming that the exclusion criteria did not systematically bias study findings (Supplementary Table S2). After applying these criteria, 340 valid responses were retained for analysis. Data Protection and confidentiality Rigorous data protection measures were implemented throughout the study to ensure participant confidentiality and comply with applicable data protection regulations. All survey responses were collected anonymously—no personally identifiable information (names, student identification numbers, email addresses, or phone numbers) was collected at any stage. The survey platform (WJX) was configured to collect only aggregate response data without linking responses to individual participants. Raw data were exported from the WJX platform and stored on secure institutional servers at West China Hospital. Access to the dataset was restricted to the research team members listed as authors, all of whom signed data confidentiality agreements. Data were identified only by automatically generated sequential numerical codes with no linkage to identifying information. During data analysis, all statistical analyses were performed on de-identified data using R software on password-protected computers in secure institutional facilities. Data storage and handling procedures were reviewed and approved by the IRB as part of the ethics application process. In accordance with institutional data retention policies, de-identified data will be securely stored for a minimum of 10 years following publication, with access restricted to the primary research team. After this retention period, data will be permanently deleted using secure data destruction protocols. No participant data will be shared with third parties. Statistical analysis Survey data were exported from the WJX platform into Microsoft Excel for initial management and subsequently analyzed using Stata software (version 17.0, StataCorp, College Station, TX, USA). Descriptive statistics were used to summarize the data: categorical variables were presented as frequencies and percentages, and continuous variables as means with standard deviations (SD). Independent samples t -tests were applied to compare continuous variables between groups, while chi-square (χ²) tests were used for categorical variables. Internal consistency reliability of multi-item scales was assessed using Cronbach’s alpha coefficient. Alpha values ≥ 0.70 were considered acceptable, ≥ 0.80 as good, and ≥ 0.90 as excellent [ 17 ]. Although a priori power calculations were not conducted for exploratory subgroup comparisons, post-hoc power analysis indicated that our sample sizes (males: 146, females: 194; academic degrees: 150, clinical degrees: 190) provide adequate statistical power (> 80%, α = 0.05) to detect small-to-medium effect sizes (Cohen’s d ≥ 0.31) for continuous variables using independent samples t-tests. For categorical variables, chi-square tests with the observed sample size provide > 99% power to detect medium effects (Cramér’s V ≥ 0.3). These thresholds are consistent with meaningful differences in educational and psychological research (Cohen, 1988). Consequently, our subgroup analyses were adequately powered to identify educationally or clinically significant differences, though very small effects (d < 0.3) may not be reliably detected. Pearson correlation coefficients were calculated to assess linear relationships between continuous variables. Correlation strength was interpreted using the following criteria: |r| < 0.3 as weak, 0.3 ≤ |r| < 0.7 as moderate, and |r| ≥ 0.7 as strong. A two-tailed p-value of < 0.05 was considered indicative of statistical significance. Results Participant demographics A total of 340 medical postgraduates participated in this study, with 146 males (42.94%) and 194 females (57.06%). The participants were coming from two institutions: West China Hospital of Sichuan University ( n = 134) and The First Affiliated Hospital of Chengdu Medical College ( n = 206). In terms of academic standing, 206 students were in Year 1, 81 in Year 2, and 53 in Year 3. The average age of participants was 23.2 years (SD = 2.2). In China, postgraduate medical education follows a dual-track system, with clinical degrees geared toward clinical practice and academic degrees aimed at research careers. Among the participants, 150 students (44.12%) were pursuing academic degrees, while 190 (55.88%) were enrolled in clinical degree programs across the two universities included in the study. Research areas included clinical medicine ( n = 264, 77.65%), nursing science ( n = 27, 7.94%), medical technology ( n = 26, 7.65%), and oral medicine/dentistry ( n = 23, 6.76%) (Table 1 ). Table 1. The demographic information of the studied medical students in this study, China 2025 Characteristics of participants The participants ( n = 340) No. % University West China Hospital of Sichuan University 134 39.41 1st Affiliated Hospital of Chengdu Medical College 206 60.59 Age (years) Mean ± SD 23.22 ± 2.20 Age Categories ≤ 23 years 181 53.24 >23 years 159 46.76 Sex Male 146 42.94 Female 194 57.06 Grade First year 206 60.59 Second year 81 23.82 Third year 53 15.59 Major Clinical medicine 264 77.65 Nursing science 27 7.94 Medical technology 26 7.65 Oral medicine/dentistry 23 6.76 Degree Academic degrees (Ph.D./M.Sc.) 150 44.12 Categories Clinical degrees (M.D./M.Med.) 190 55.88 Open in a new tab Abbreviations : SD Standard Deviation, Ph.D. Doctor of Philosophy, M.Sc. Master of Science, M.D. Doctor of Medicine, M.Med. Master of Medicine Awareness, usage patterns and purposes of AI chatbot We found that 82.89% of the students demonstrated a solid understanding of and familiarity with general AI concepts. The majority (253 students, 74.6%) reported knowledge of machine learning or deep learning, while 210 students (61.95%) were familiar with natural language processing (NLP). Notably, only 76 students (22.35%) indicated familiarity with convolutional neural networks (CNNs). In terms of generative AI tool usage in daily academic activities, DeepSeek was the most widely used, with 90.86% of students reporting regular use. Over half of the respondents (55.16%) also used ChatGPT. Other AI tools were used to a lesser extent, including Gemini (9.14%), MetaAI (5.9%), Grok (4.13%), Co-Pilot (4.13%), and Claude (3.24%). When asked to identify their preferred AI chatbot, 246 (72.35%) of students selected DeepSeek as their top choice as illustrated in Fig. 1 . Fig. 1. Open in a new tab Distribution of preferred AI chatbots among postgraduate medical students at two colleges in China. Postgraduate medical students ( n =340) were asked to select their most preferred AI chatbot from available options. DeepSeek was selected by 72.35% of participants as their top choice Regarding the frequency of AI chatbot usage, 33.53% of students reported using AI tools most of the time, while 45.00% used them often. An additional 18.53% used AI tools barely, and 2.94% never used AI chatbots. In terms of application, the most common uses among medical students included literature research and review (61.47%), academic question answering and research inspiration (55.88%), course study or exam preparation (55.00%), and English translation or editing (54.12%). A substantial proportion also employed AI tools for experimental design or data analysis (48.53%) and academic writing (43.24%). A smaller percentage (14.71%) utilized AI to assist with grant or project proposal preparation. (Fig. 2 a). Fig. 2. Open in a new tab Distribution of AI chatbot applications among 340 postgraduate medical students. a Use across academic and research activities. b Use in specific sections of academic writing. Percentages represent proportion of students using AI for each purpose Further analysis examined the specific sections of academic writing where students applied generative AI. The highest reported usage was for drafting introductions (47.94%), followed by methods Sect. (46.76%) and abstract writing (46.18%). Usage was also notable in discussion sections and grammar refinement (44.71% in each), and results Sect. (39.12%). Citation suggestions were less commonly supported by AI tools, with 31.76% of students reporting use in this area (Fig. 2 b). Among students who used AI chatbots, 99.11% reported overall satisfaction, with a mean satisfaction score of 4.4 on a 5-point Likert scale (1 = very dissatisfied, 5 = very satisfied). Regarding the perceived suitability of AI in different types of medical courses, 46.45% of students believed that AI is best suited to assist with medical literature reading. Additionally, 23.08% indicated that AI is applicable to basic medical sciences, such as anatomy and pathology. A smaller proportion (13.01%) considered AI suitable for research methodology. Notably fewer students felt that AI was well suited for doctor–patient communication simulation (10.06%) and clinical skills training (7.40%) (Fig. 3 ). Fig. 3. Open in a new tab Student perceptions of the most suitable medical course types for AI integration. The pie chart illustrates students’ views on the most appropriate applications of AI, including medical literature reading, basic medical sciences, research methodology, doctor–patient communication simulation, and clinical skills training Perceptions of AI chatbots To explore how medical students perceive the value and impact of AI chatbots in their academic education and experience, this study evaluated participants’ views on the effectiveness of these tools in supporting both educational and research activities. Specifically, students were asked to evaluate how AI chatbots had influenced their performance in tasks such as information retrieval, academic writing, and knowledge synthesis. A majority of students (70.80%) reported that AI significantly improved the efficiency of information retrieval. Additionally, 64.31% indicated that AI contributed to more precise language use in writing. Over half of the respondents agreed that AI enhanced writing quality (56.05%) and facilitated a clearer structure and organization of ideas (58.11%). Notably, only three students (0.88%) believed that AI had not improve their writing efficiency. Students also reported perceived improvements in specific skills following AI tool usage. Notably, 78.10% believed their information retrieval efficiency had improved, 50.59% felt they had enhanced critical thinking abilities with AI assistance, 57.99% reported increased confidence in clinical decision-making, and 47.34% believed AI had helped improve long-term knowledge retention. Attitudes, concerns and expectations toward AI in medical education Overall, medical students exhibited a positive attitude toward the integration of AI in medical education. In this study, 84.03% of students agreed that it is essential to incorporate AI into the medical curriculum. More than half (51.19%) believed that AI-driven learning would eventually replace traditional classroom instruction. A substantial majority (85.51%) indicated that integrating AI with clinical medical education would improve overall learning efficiency. Furthermore, 82.54% of students felt that AI enhances personalized instruction and enables timely feedback. Regarding AI-based scoring systems, 68.04% believed they could improve the accuracy and fairness of academic assessments. Additionally, 73.01% agreed that AI-powered clinical simulation provides a safe environment for developing clinical skills. In the context of simulated diagnosis and case-based teaching, 77.81% of students reported that AI tools helped them identify key information and make preliminary judgments more quickly and accurately. In terms of future plans, 83.73% of students indicated they would apply AI tools in future teaching or research, and 77.22% planned to use AI to assess their own and their peers’ learning progress. A further 79.29% expressed willingness to recommend AI tools to classmates or mentors, and 78.99% expressed interest in participating in AI-related medical research projects (Fig. 4 ). Fig. 4. Open in a new tab Student attitudes toward AI integration in medical education ( n =340) assessed using 5-point Likert scales. Responses are displayed as percentages for each attitude statement, categorized as Strongly Disagree, Disagree, Neutral, Agree, and Strongly Agree Despite these benefits, students expressed several concerns regarding the limitations of AI in medical education. Over one-third (36.28%) were worried that AI may reduce opportunities for communication and collaboration among peers. Concerns were also raised about an overemphasis on data skills at the expense of soft skills (39.23%), diminished personalized attention from instructors (33.03%), and the over-quantification of learning outcomes (37.17%). Additionally, students reported worries about the accuracy of AI outputs (40.41%), potential for academic misconduct such as plagiarism (49.55%), data privacy and security issues (45.43%), and reduced critical thinking development (37.75%) (Fig. 5 ). Fig. 5. Open in a new tab Student concerns regarding AI use in medical education assessed using 5-point Likert scales. Responses show levels of concern (Very Concerned, Concerned, Neutral, Unconcerned, Not at All Concerned) for each potential risk factor Finally, students identified several key areas where they hoped AI chatbots would improve. These included enhanced accuracy and professionalism of responses (89.68%), better support for staying updated with research trends and literature (75.22%), improved contextual understanding in multi-turn conversations (72.27%), stronger data privacy and security protection (63.42%), and faster, more stable response performance (55.16%). Reliability of multi-item scales All multi-item scales demonstrated good to excellent internal consistency. The Attitudes toward AI Integration scale (7 items) yielded a Cronbach’s α of 0.886 (95% CI: 0.868–0.905). The Behavioral Intentions scale (4 items) showed excellent reliability (α = 0.922, 95% CI: 0.908–0.935). Both concern scales demonstrated excellent internal consistency: General Concerns (4 items, α = 0.901, 95% CI: 0.883–0.918) and Medical Education Concerns (4 items, α = 0.914, 95% CI: 0.899–0.929). The Perceived Ability Changes scale (4 items) showed good reliability (α = 0.858, 95% CI: 0.833–0.883). These reliability coefficients indicate that all scales had sufficient internal consistency for research purposes (Supplementary Table 1). Subgroup analyses and Pearson correlation analysis Subgroup analyses were performed to compare awareness of AI chatbots, usage patterns, perceptions, attitudes, and concerns across gender and degree types. Usage patterns differed by gender, with female students more likely to report frequent AI use compared to males ( p = 0.025, Cohen’s d = 0.25). Female students also expressed greater concerns regarding the risks of AI use in general ( p < 0.001, Cohen’s d = 0.45) as well as risks specific to medical education ( p < 0.001, Cohen’s d = 0.38). No significant differences were observed between male and female students in terms of awareness scores, educational purposes of AI use, or overall attitudes toward AI. When stratified by degree program, clinical-track postgraduate students who have clinical programs demonstrated higher awareness of AI compared with their counterparts in academic-track postgraduate students who have academic programs ( p = 0.001, Cohen’s d = 0.34). Additionally, clinical-track students reported higher overall satisfaction with AI chatbots ( p = 0.003, Cohen’s d = 0.31). In contrast, no significant differences were found between the two groups regarding usage frequency, educational purposes of AI, or concern levels. Pearson correlation analysis was conducted to examine the relationships among students’ awareness, frequency of AI usage, perception (helpfulness rating), attitudes, and concerns regarding AI in medical education. As shown in Fig. 6 , awareness of AI was positively associated with both usage frequency ( r = 0.273, p < 0.001) and perception score of AI ( r = 0.393, p < 0.001). Usage frequency also showed a moderate positive correlation with perception score ( r = 0.402, p < 0.001), suggesting that students who use AI tools more frequently tend to perceive them as more helpful. Fig. 6. Open in a new tab Pearson correlation matrix of key variables related to AI adoption in medical education ( n =340). Correlation coefficients are displayed with significance levels (* p <0.05, ** p <0.01, *** p <0.001) Attitudes toward AI in education and in academic work were significantly correlated with both perception and usage. Notably, attitudes toward AI in education and academic work were strongly associated ( r = 0.661, p < 0.001), suggesting consistent positive views across educational and academic contexts. Both attitude domains also showed weak to moderate positive correlations with perception score (education: r = 0.289, p < 0.001; academic work: r = 0.350, p < 0.001). In contrast to the positive associations among awareness, usage, and attitudes, concern-related variables exhibited weaker or negligible correlations with awareness and usage. Specifically, concerns about AI usage and concerns in medical education showed no significant correlation with awareness ( r = 0.030 and − 0.003, respectively, both p > 0.05) or usage frequency ( r = − 0.080 and − 0.104, respectively, both p > 0.05), suggesting that higher exposure to AI does not necessarily reduce levels of concern. However, the two concern domains were strongly positively correlated with each other ( r = 0.783, p < 0.001), indicating that individuals who are more concerned about AI usage in general also tend to express greater concern about its impact in medical education. Additionally, attitudes toward AI in both education and academic work showed moderate positive correlations with both concern variables (r ranging from 0.199 to 0.310, all p < 0.001), implying that even those with favorable attitudes may still harbor significant concerns about its implementation and potential risks. Discussion In this cross-sectional study of Chinese postgraduate medical students, we found high levels of awareness of generative AI and widespread use of chatbots for academic tasks, particularly scientific writing, literature review, and language refinement. From a technology-adoption perspective, these patterns suggest that students perceive generative AI chatbots as offering clear performance advantage and ease of integration into existing workflows—the core drivers of adoption in the Technology Acceptance Model (TAM) [ 18 – 20 ]. Our findings align with emerging evidence that learners embrace AI tools when they perceive them as efficient, accessible, and aligned with academic needs [ 20 ], demonstrating high perceived usefulness and adequate perceived ease of use—TAM’s foundational constructs for predicting technology acceptance [ 21 ]. AI technologies are increasingly recognized for their ability to transform medical education by providing personalized instruction, adapting content to individual learning needs, and supporting remote education through interactive, intelligent platforms. These capabilities contribute to greater learning efficiency and improve educational quality [ 12 ].To situate our findings within the global landscape of AI adoption in medical education, we found that 82.9% of Chinese postgraduate students demonstrated solid understanding of AI concepts and 78.5% reported frequent or most-time usage of AI chatbots represents notably higher adoption compared to the pan-European study by Busch et al. [ 7 ] reported 58.3% usage among medical students across 192 faculties, suggesting China-specific factors: rapid COVID-19-accelerated educational digitalization, intense publication pressures (stringent journal publication requirements as graduation prerequisites), and availability of sophisticated Chinese-language tools like DeepSeek. Moreover, sophisticated Chinese-language AI tools like DeepSeek reduce language barriers constraining adoption in non-English-speaking countries where students must rely on English-optimized tools. While ChatGPT dominated Western studies (> 60% adoption) [ 22 ], our participants preferred DeepSeek (72.4%), indicating linguistic compatibility and cultural localization significantly influence tool selection—patterns insufficiently explored in Western-centric literature [ 23 ]. Usage purposes showed universal patterns (literature review 61.5%, exam preparation 55.0%, writing 43.2%) mirroring international studies of health professions students’ GenAI use that report widespread utilization for academic tasks such as information retrieval and writing support. Surveys of medical and allied health students in the U.S. and Canada found that a majority reported using generative AI tools for learning and summarizing content, with substantial proportions citing academic purposes including writing and guideline summarization [ 24 ]. However, the relatively high experimental design/data analysis use (48.5%) observed in our sample likely reflects China’s dual-track medical education system and associated academic pressures, where postgraduate students are expected to engage substantively in research and publication activities that are less emphasized in some Western curricula. From a Diffusion of Innovation (DOI) perspective, the relative advantage of AI tools in accelerating research productivity may be amplified in this high-pressure “publish or perish” environment, fostering rapid adoption among trainee researchers [ 20 ]. Our postgraduate participants appear to be advanced adopters compared to studies reporting cautious use [ 12 ], with positive correlations among awareness ( r = 0.273), usage frequency, and perception scores ( r = 0.402) confirming TAM predictions that perceived usefulness drives sustained adoption [ 20 ]. Compared with previous studies of medical and health professions students, which have reported variable levels of AI literacy and more cautious use of generative AI tools [ 25 , 26 ], the postgraduate students in our sample appear to be relatively advanced adopters. Their frequent use of chatbots for drafting manuscripts, summarizing evidence, and refining academic English suggests that generative AI has already become embedded in the research culture of early-career clinicians and scientists. Similar to TAM-based studies of large language models and AI assistants in clinical medicine and nursing education [ 18 , 27 ], our results underscore the central importance of perceived usefulness: students who believe that AI tools improve efficiency, clarity, or confidence in academic work are those who use them most extensively. Beyond academic writing, the programming capabilities of generative AI have also been recognized as valuable in fields such as bioinformatics education. Prior work demonstrated that AI chatbots can assist beginners in coding by generating accurate scripts, identifying errors, and providing real-time corrections and explanations, thereby improving both efficiency and motivation [ 28 ]. Postgraduate medical education in China has two types of medical programs which differs from many other countries [ 29 ]. Clinical-track students are required to engage in research, often with publication demands, while academic-track students are simultaneously expected to balance intensive scientific work with some level of clinical exposure [ 30 ]. This dual emphasis creates unique pressures and learning needs, making it essential to investigate how Chinese postgraduate medical students adopt generative chatbots, assess their attitudes and concerns, and explore their behavioral intentions toward AI across both academic and clinical pathways. The subgroup analyses showed that clinical-track postgraduate students reported significantly higher awareness scores and perceived greater benefits from AI compared with academic-track students. This pattern is consistent with DOI, which highlights how exposure to innovation-relevant tasks and social networks can accelerate adoption [ 31 ]. Clinical-track students encounter AI tools more directly in their training environments—through clinical decision support systems, diagnostic imaging applications, and patient management platforms—thereby experiencing greater relative advantage and observability of AI benefits in real-world practice settings. By contrast, academic-track students, who focus primarily on laboratory research and traditional scientific methodologies, may have less frequent exposure to applied AI tools, contributing to lower awareness despite their strong research capabilities. These findings underscore the importance of tailoring AI education to the specific contexts and needs of both clinical and academic training pathways [ 32 , 33 ]. Despite enthusiasm, students expressed persistent concerns about accuracy (40.4%), plagiarism (49.6%), and privacy (45.4%). These map onto perceived risk and trust in extended TAM/UTAUT models [ 34 ]. Critically, concerns regarding content accuracy, ethical use, and overreliance on AI tools remained independent of awareness and usage, suggesting increased exposure alone does not reduce skepticism—sustained responsible use requires addressing ethical concerns and establishing institutional trust [ 35 , 36 ]. The gender difference (females showing higher usage yet greater concerns) suggests “cautious adopters” who actively use AI while remaining vigilant—a nuanced profile highlighting that high perceived usefulness can coexist with high perceived risk [ 37 ]. Our findings indicate curricula focusing solely on “how to use” AI are insufficient; critical appraisal, policy understanding, and academic integrity skills are equally essential. The outcomes of our study hold the view that students’ views on AI were associated with their familiarity with AI theory and perceived advantages such as improved learning efficiency. The correlation analysis of our study reveals that medical students’ attitudes toward AI are significantly shaped by their awareness and usage frequency of AI tools. The presence of concern appears to function independently of awareness or usage. The lack of significant correlations between concern variables and AI exposure suggests that simply interacting more with AI does not diminish skepticism or unease. Previous research has shown that higher AI literacy—defined as both conceptual understanding and practical familiarity—correlates with more favorable attitudes among medical students and healthcare professionals [ 11 ]. However, increased exposure alone may not fully mitigate ethical concerns, such as data privacy, algorithmic bias, and loss of human interaction, which often persist even among AI users [ 38 ]. This aligns with findings from other studies indicating that while generative AI tools can enhance performance in tasks like writing and diagnosis simulation, skepticism remains, particularly around over-reliance and the erosion of critical thinking [ 39 ]. The coexistence of enthusiasm and caution in medical students reflects a nuanced stance: they are willing to adopt AI when its value is evident but remain alert to its limitations and long-term implications. Addressing this duality may require not only technical training but also structured dialogue around the role of AI in professional identity, ethics, and patient safety. Building on TAM/DOI insights and our findings, we propose targeted strategies for responsible integration of generative AI chatbots into postgraduate medical education across three domains. First, students are already using AI tools ad-hoc without structured guidance. Medical curricula should include mandatory AI literacy modules covering ethical use, critical evaluation of outputs, and practical demonstrations, followed by embedded AI practice in existing courses such as Scientific Writing and Evidence-Based Medicine. This aligns with scoping reviews that emphasize the need for structured AI curriculum frameworks and tailored educational programs in medical education to prepare learners for AI-augmented practice [ 40 ].Second, successful integration requires systematic training for educators who often lack experience with AI. Faculty development should include workshops on AI fundamentals, pedagogical integration of AI tools, and ethical facilitation, supported by institutional task forces and shared resources. Reviews of AI in medical education highlight the necessity of faculty training and institutional support to translate high student interest into effective teaching implementation [ 41 ]. Third, to address concerns such as academic integrity, institutions should define clear guidelines on permissible AI use and integrate AI-resistant and AI-enhanced assessment components (e.g., reflective portfolios, oral defenses, and AI-critique assignments). Innovations in assessment design that responsibly incorporate AI tools have been identified as a priority area for authentic evaluation in medical education [ 42 ].These evidence-based strategies provide actionable steps for embedding AI literacy, supporting faculty capacity, and reforming assessment, thereby advancing responsible and pedagogically sound AI integration in postgraduate medical training. This study has several limitations that should be considered when interpreting the findings. First, convenience sampling from two Chengdu universities may not represent broader Chinese postgraduate populations. Regional disparities in China are substantial: eastern coastal cities typically have greater access to advanced technology and international educational resources, while central and western regions may face infrastructure limitations. Second, our study’s cross-sectional design precents assessment of causality or temporal changes in AI adoption patters. Third, the reliance on self-reported data introduces the possibility of response bias, including recall errors and social desirability effects, which may have affected the accuracy of reported usage patterns and attitudes. International comparative studies using harmonized instruments would systematically distinguish universal from China-specific patterns. Future research should employ multi-center longitudinal designs tracking how TAM constructs evolve with AI experience and institutional support maturation. In conclusion, our results position Chinese postgraduate medical students as pivotal in early generative AI diffusion within medical education, extending prior undergraduate-focused work [ 12 ]. From a TAM perspective, the high reported usage and satisfaction scores reflect strong perceived usefulness and adequate perceived ease of use—foundational prerequisites for technology acceptance [ 6 ]. From a DOI perspective, the rapid adoption of tools like DeepSeek and ChatGPT suggests these innovations have crossed the “early adopter” phase in this population, driven by clear relative advantage in academic productivity and compatibility with research workflows [ 20 ]. However, persistent concerns despite high usage indicate medical educators face a critical challenge: facilitating responsible diffusion rather than merely promoting availability. Our findings highlight urgent need to transition from ad-hoc, student-driven use toward structured, curriculum-integrated approaches explicitly addressing opportunities (writing support, self-directed learning) and risks (hallucinations, integrity erosion). The actionable recommendations we provide for curriculum, faculty development, and assessment offer a roadmap balancing innovation with integrity, efficiency with critical thinking, and technology with humanistic values. Institutions proactively implementing such frameworks will better prepare postgraduate students for AI-augmented practice while addressing legitimate stakeholder concerns. This study provides preliminary evidence from two well-resourced Chengdu institutions, establishing foundation for broader longitudinal investigations of AI’s long-term educational impact in Chinese medical education. Supplementary Information Supplementary Material 1. (17.9KB, docx) Acknowledgements None. Authors’ contributions Conceptualization: Lijun Zhao. Data curation: Lijun Zhao, Qianqian Han and Peijuan Li. Formal analysis: Lijun Zhao. Writing – original draft: Lijun Zhao. Writing – review and editing: Hua Dai and Xuegui Ju. Supervision: Hua Dai and Xuegui Ju. All authors actively participated in the research process, made substantial contributions to manuscript revisions, and carefully reviewed and approved the final version. All authors read and approved the final manuscript. Funding This study was supported by the following funding sources: Funder One, Sichuan University Higher Education Teaching Reform Project (Phase XI), Grant/Award Number: SCU11144 (Recipient: Lijun Zhao); Funder Two, Chengdu Science and Technology Bureau, Technological Innovation and Research & Development Project, Grant/Award Number: 2024-YF05-00409-SN (Recipient: Lijun Zhao); Funder Three, Science & Technology Department of Sichuan Province, Science and Technology Training Program, Industry Science Popularization Capacity Enhancement Project, Grant/Award Number: 2025JDKP0031 (Recipient: Lijun Zhao); Funder Four, West China Hospital, Sichuan University, 1·3·5 project for disciplines of excellence–Clinical Research Fund, Grant/Award Number: 2024HXFH020 (Recipient: Lijun Zhao); Funder Five, Chengdu Medical College, Education Reform Research Project, Grant/Award Number: JG201905 (Recipient: Xuegui Ju). The funding bodies had no role in the design of the study, data collection, analysis, interpretation of data, or in writing the manuscript. Data availability The datasets generated during the current study are not publicly available due to privacy restrictions concerning participant information but are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was approved by the Institutional Review Board (IRB) of West China Hospital in Sichuan University (Approval No. 2023 − 1066). Participation was voluntary, and all participants were informed about the purpose of the study, procedures, and confidentiality measures. Written informed consent was obtained from all participants prior to data collection. Consent for publication Not applicable. This manuscript does not contain any individual person’s data in any form (including images, videos, or case reports). Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Hua Dai, Email: [email protected]. Xuegui Ju, Email: [email protected]. References 1. Liu X, et al. A comparison of deep learning performance against health-care professionals in detecting diseases from medical imaging: a systematic review and meta-analysis. Lancet Digit Health. 2019;1(6):e271–97. [ DOI ] [ PubMed ] [ Google Scholar ] 2. Topol EJ. High-performance medicine: the convergence of human and artificial intelligence. Nat Med. 2019;25(1):44–56. [ DOI ] [ PubMed ] [ Google Scholar ] 3. World Health Organization. Ethics and governance of artificial intelligence for health: guidance on large multi-modal models. Geneva: World Health Organization; 2024. Available from: https://www.who.int/publications/i/item/9789240084759 . 4. Abdelhafiz AS, et al. Knowledge, Perceptions and Attitude of Researchers Towards Using ChatGPT in Research. J Med Syst. 2024;48(1):26. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Tolsgaard MG, et al. The fundamentals of Artificial Intelligence in medical education research: AMEE Guide 156. Med Teach. 2023;45(6):565–73. [ DOI ] [ PubMed ] [ Google Scholar ] 6. Lee JWY, Tan JY, Bello F. Technology Acceptance Model in Medical Education: Systematic Review. JMIR Med Educ. 2025;11:e67873. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Busch F, et al. Global cross-sectional student survey on AI in medical, dental, and veterinary education and practice at 192 faculties. BMC Med Educ. 2024;24(1):1066. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Alammari D, et al. Assessing medical students’ attitudes, performance, and usage of ChatGPT in Jeddah, Saudi Arabia. Front Artif Intell. 2025;8:1577911. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Kung TH, 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 ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Abdelhafiz AS, et al. Medical students and ChatGPT: analyzing attitudes, practices, and academic perceptions. BMC Med Educ. 2025;25(1):187. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Sallam M. ChatGPT utility in healthcare education, research, and practice: systematic review on the promising perspectives and valid concerns. Healthc (Basel). 2023;11(6):887. [ DOI ] [ PMC free article ] [ PubMed ] 12. Duan S, et al. Integrating AI in medical education: a comprehensive study of medical students’ attitudes, concerns, and behavioral intentions. BMC Med Educ. 2025;25(1):599. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Eysenbach G. Improving the quality of Web surveys: the Checklist for Reporting Results of Internet E-Surveys (CHERRIES). J Med Internet Res. 2004;6(3):e34. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. von Elm E, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–9. [ DOI ] [ PubMed ] [ Google Scholar ] 15. Ghanem OA, et al. Medical students’ knowledge, attitudes, and practices toward generative artificial intelligence in Egypt 2024: a Cross-Sectional study. BMC Med Educ. 2025;25(1):790. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Rodman A, et al. Using Generative Artificial Intelligence in Medical Education. Acad Med. 2025;100(2):250. [ DOI ] [ PubMed ] [ Google Scholar ] 17. Tavakol M, Dennick R. Making sense of Cronbach’s alpha. Int J Med Educ. 2011;2:53–5. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Liu F, et al. Assessing clinical medicine students’ acceptance of large language model: based on technology acceptance model. BMC Med Educ. 2024;24(1):1251. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Zhang R, Wang J. Perceptions, adoption intentions, and impacts of generative AI among Chinese university students. Curr Psychol. 2025;44(11):11276–95. [ Google Scholar ] 20. Tao W, Yang J, Qu X. Utilization of, Perceptions on, and Intention to Use AI Chatbots Among Medical Students in China: National Cross-Sectional Study. JMIR Med Educ. 2024;10:e57132. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Cook DA, et al. Virtual Patients Using Large Language Models: Scalable, Contextualized Simulation of Clinician-Patient Dialogue With Feedback. J Med Internet Res. 2025;27:e68486. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Morell-Mengual V, et al. Characteristics, motivations and attitudes of students using ChatGPT and other language model-based chatbots in higher education. Educ Inform Technol. 2025;30(15):22257–74. [ Google Scholar ] 23. Barnes AJ, Zhang Y, Valenzuela A. AI and culture: Culturally dependent responses to AI systems. Curr Opin Psychol. 2024;58:101838. [ DOI ] [ PubMed ] [ Google Scholar ] 24. Harris EC, et al. Surveying Health Professions Students’ Use of Generative AI Applications. Med Sci Educ. 2025;35(4):1907–11. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Hu N, et al. Status and perceptions of ChatGPT utilization among medical students: a survey-based study. BMC Med Educ. 2025;25(1):831. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 26. Gualda-Gea JJ, et al. Perceptions and future perspectives of medical students on the use of artificial intelligence based chatbots: an exploratory analysis. Front Med (Lausanne). 2025;12:1529305. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Labrague LJ, Al M, Harrasi. Nursing students’ perceptions of artificial intelligence (AI) using the technology acceptance model: A systematic review. Teach Learn Nurs. 2025;20(3):274–82. [ Google Scholar ] 28. Shue E, et al. Empowering beginners in bioinformatics with ChatGPT. Quant Biol. 2023;11(2):105–8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Liu X, et al. Medical Education Systems in China: Development, Status, and Evaluation. Acad Med. 2023;98(1):43–9. [ DOI ] [ PubMed ] [ Google Scholar ] 30. Tie H, Luo Y, Chen D. Thinkings on the reform of medical education system in China. Med Educ Online. 2024;29(1):2302677. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Putteeraj M, et al. Assessing E-Health adoption readiness using diffusion of innovation theory and the role mediated by each adopter’s category in a Mauritian context. Int Health. 2022;14(3):236–49. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Li Q, Qin Y. AI in medical education: medical student perception, curriculum recommendations and design suggestions. BMC Med Educ. 2023;23(1):852. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Banerjee M, et al. The impact of artificial intelligence on clinical education: perceptions of postgraduate trainee doctors in London (UK) and recommendations for trainers. BMC Med Educ. 2021;21(1):429. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 34. Lai CY, et al. Integrating the adapted UTAUT model with moral obligation, trust and perceived risk to predict ChatGPT adoption for assessment support: A survey with students. Computers Education: Artif Intell. 2024;6:p100246. [ Google Scholar ] 35. Wenhang S, Shenting Z, Shaohua C. Prospects for the development of medical professionalism education in the AI perspective: A qualitative study of Chinese postgraduate medical students’ written reflections. Med Teach. 2025;47(9):1535–43. [ DOI ] [ PubMed ] [ Google Scholar ] 36. Sharma S, Saboo N, Baig VN. Exploring Undergraduate Medical Students’ Perspectives towards Artificial Intelligence in Healthcare: A Qualitative Study from India. Jma j. 2025;8(3):718–29. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 37. Homko CJ, et al. Gender differences in cardiovascular risk factors and risk perception among individuals with diabetes. Diabetes Educ. 2010;36(3):483–8. [ DOI ] [ PubMed ] [ Google Scholar ] 38. Malerbi FK, et al. Digital Education for the Deployment of Artificial Intelligence in Health Care. J Med Internet Res. 2023;25:e43333. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Preiksaitis C, Rose C. Opportunities, Challenges, and Future Directions of Generative Artificial Intelligence in Medical Education: Scoping Review. JMIR Med Educ. 2023;9:e48785. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 40. Tolentino R, et al. Curriculum Frameworks and Educational Programs in AI for Medical Students, Residents, and Practicing Physicians: Scoping Review. JMIR Med Educ. 2024;10:e54793. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. Shaw K, Henning MA, Webster CS. Artificial Intelligence in Medical Education: a Scoping Review of the Evidence for Efficacy and Future Directions. Med Sci Educ. 2025;35(3):1803–16. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. Fatima SS, Sheikh NA, Osama A. Authentic assessment in medical education: exploring AI integration and student-as-partners collaboration. Postgrad Med J. 2024;100(1190):959–67. [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary Material 1. (17.9KB, docx) Data Availability Statement The datasets generated during the current study are not publicly available due to privacy restrictions concerning participant information but are available from the corresponding author on reasonable request. Articles from BMC Medical Education are provided here courtesy of BMC ACTIONS View on publisher site PDF (2.3 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top