Quick Buzz Feed
This cross-sectional study assessed the readiness and perceptions of undergraduate medical students towards artificial intelligence (AI) at a tertiary care teaching institute in India. It found that students demonstrated an average level of AI readiness and generally mixed-to-positive perceptions. Prior AI training was significantly associated with readiness in the ethics domain, while age and academic year showed only a weak, non-significant correlation. The study highlights the need for structured AI education, particularly in ethical principles, to prepare future physicians for AI integration in healthcare.
Artificial intelligence (AI) is rapidly transforming healthcare, with applications ranging from diagnostics to individualized treatment planning, making its integration into medical education crucial. Future physicians need comprehensive AI literacy, encompassing conceptual knowledge, practical skills, and ethical reasoning, to effectively evaluate AI tools, address algorithmic biases, safeguard patient data, and critically engage with ethical considerations. International studies indicate that while medical students recognize AI's educational potential, they often report deficiencies in practical competence and ethical understanding. This study aims to address a regional evidence gap in India by assessing AI readiness and perceptions among undergraduate Bachelor of Medicine, Bachelor of Surgery (MBBS) students in Gujarat. It also examines how this readiness associates with sociodemographic characteristics, prior AI training, and AI tool utilization patterns.
This cross-sectional study was conducted between September and November 2025 at Shrimad Rajchandra Sarvamangal Hospital & C U Shah Medical College, Surendranagar, Gujarat. A universal (census) sampling approach invited all 400 eligible undergraduate MBBS students, with 310 completing the survey (77.5% response rate). Inclusion criteria were current enrollment and electronic informed consent, while absence or lack of consent led to exclusion. Ethical approval was obtained from the Institutional Ethics Committee. Data were collected via a self-administered Google Forms questionnaire, comprising four sections: sociodemographic information, AI tool exposure and utilization, the 22-item Medical Artificial Intelligence Readiness Scale for Medical Students (MAIRS-MS) (administered to 238 AI users), and a validated 10-item questionnaire assessing AI perception (administered to all 310 participants). Statistical analysis involved descriptive statistics (means, standard deviations, frequencies, percentages), Pearson's chi-square test, Fisher's exact test, and Spearman's rank correlation coefficient. AI readiness was categorized into poor (≤66), average (67-78), or good (≥79) based on tertiles of the observed MAIRS-MS score distribution, with a two-sided p-value <0.05 considered statistically significant.
The study included 310 participants, with a mean age of 20.03 ± 1.65 years, and 58.71% were female. Most (77.42%) resided in urban areas. Among them, 238 (76.77%) had used AI for academic purposes in the past year, predominantly using free tools (95.38%), with ChatGPT (92.86%) being the most popular. Only 26.47% had received formal AI training, and social media (80.67%) was the primary source of information. Key difficulties reported included the need for paid versions (66.81%), faulty data (30.25%), and poor reliability of outputs (28.15%). The mean MAIRS-MS score for AI users was 71.75 ± 14.09, classifying 71.0% as having average AI readiness. Subscale analysis showed higher mean scores for cognition (25.23 ± 5.57) and ability (26.66 ± 5.77) compared to vision (9.82 ± 2.36) and ethics (10.04 ± 2.28). Previous AI training significantly associated with the ethics domain of AI readiness (χ²=6.33, p=0.042), but no other sociodemographic or AI-related variables showed significant associations with readiness. Age and academic year had extremely weak, non-significant positive correlations with total AI readiness. Participants largely perceived AI as useful for teaching (53.23%), assignment preparation (52.26%), self-learning (53.55%), understanding complex concepts (54.84%), and clinical case scenarios (51.93%). However, concerns were substantial regarding misleading information (48.70%), negative effects on clinical skills and critical thinking (48.71%), and data privacy (42.90%), with a considerable proportion of students remaining neutral on many aspects.
This study revealed an average level of AI readiness among undergraduate medical students, accompanied by mixed-to-positive perceptions regarding AI's role in medical education. Consistent with other research, students demonstrated higher readiness in AI cognition and ability domains but lower in vision and ethics, indicating a gap in understanding AI's broader implications and ethical responsibilities. Despite widespread AI tool usage (76.77%), primarily free tools like ChatGPT and information from social media, only a quarter had formal AI training. This reliance on informal learning suggests a potential for misconceptions and a lack of rigorous, evidence-informed understanding. Students cited practical barriers like the need for paid subscriptions, faulty data, and unreliable outputs, underscoring the need for structured educational interventions. While AI was seen as beneficial for various educational purposes, concerns about misinformation, negative impacts on clinical skills, and data privacy were prominent, reflecting a nuanced perspective developed through self-directed use. Crucially, prior AI training was significantly associated with higher readiness in the ethics domain, supporting the integration of ethical principles into AI education. The absence of significant correlations between readiness and sociodemographic factors or frequency of AI tool use suggests that formal, structured instruction, rather than mere exposure, is essential for developing a deep and responsible understanding of AI in medical practice.
This study concludes that undergraduate medical students at a tertiary care teaching institute in Gujarat generally exhibit an average level of AI readiness and mixed-to-positive perceptions toward artificial intelligence, with a notable proportion remaining neutral on various aspects. The only variable significantly linked to improved AI readiness was previous formal AI training, specifically within the ethics domain. Conversely, sociodemographic factors like age and academic year showed only an extremely weak and non-significant correlation with AI readiness. Students widely recognized AI's utility in educational contexts such as teaching, assignment preparation, self-learning, understanding complex concepts, and clinical case scenarios. However, they also expressed considerable concerns regarding the provision of misleading information, potential adverse effects on clinical skills and critical thinking, and issues related to data privacy. These findings strongly advocate for the integration of structured AI education into undergraduate medical curricula, with a particular emphasis on fostering ethical principles, to adequately prepare future healthcare professionals for the appropriate and responsible use of AI in medicine.