Artificial intelligence knowledge, perceptions, and educational needs among physician associate students in Kerala, India: a cross-sectional study

Abstract

Artificial intelligence (AI) is increasingly being integrated into healthcare through applications in clinical decision support, diagnostic imaging, and digital health. Despite growing interest in AI among healthcare students, evidence regarding Physician Associate (PA) students, particularly in low- and middle-income countries, remains limited. This exploratory study assessed AI-related knowledge, perceptions, ethical concerns, and educational needs among PA students in India. A cross-sectional survey, conducted and reported in accordance with the STROBE checklist for cross-sectional studies, was carried out between June and August 2025 among PA students enrolled at a tertiary academic institution in Kerala, India (Institutional Ethics Committee approval no. (ECR129/inst/KL/2013/RR-24). All students from first year through internship were eligible. Data were collected using a questionnaire adapted from a previously validated instrument, which underwent expert content-validity review and pilot testing prior to administration. Descriptive statistics summarised responses, and chi-square tests examined associations between participant characteristics and selected perception items. Of 80 eligible students, 70 were included in the final analysis (87.5% of the eligible cohort). The mean age was 20.64 ± 2.46 years, and 81.4% were female. Only 17.1% of participants considered themselves knowledgeable about AI, while 42.9% reported partial knowledge and 40.0% reported no knowledge. All participants agreed that AI could assist physician associates rather than replace them, while 35.7% expressed concern that AI could reduce future employment opportunities. Ethical concerns included potential reduction in the humanistic aspects of healthcare (65.7%), negative effects on clinician–patient relationships (55.7%), and confidentiality concerns (48.6%). All participants supported structured AI training as part of their medical education. Preclinical students were significantly more likely than clinical-year students to perceive AI as a threat to employment opportunities (46.9% vs. 9.5%; p  = 0.005). PA students demonstrated generally favourable perceptions of AI and strong demand for formal AI education, but reported limited knowledge and confidence. The coexistence of positive attitudes, uncertainty, and ethical concerns highlights the need for structured AI education that integrates foundational AI literacy, clinical applications, critical appraisal, and ethical considerations. As an exploratory single-institution study, these findings are hypothesis-generating. Larger multicentre studies using objective measures of AI competency are warranted.