Studying with GenAI: cross-sectional study on usage patterns, needs, competencies, and ethical perspectives of medical informatics students

Background Generative AI use (GenAI) by students is transforming education and professional practice, necessitating literacy training. Comprehensive data on usage, competencies, needs and ethics, especially in the Netherlands, is scarce. To inform future literacy education, this study examines usage frequency and type, self-assessed competencies, ethical attitudes, and perceived career impact among Medical Informatics students at the University of Amsterdam. Methods From September 2024 to February 2025, we conducted a cross-sectional survey among Bachelor and Master students. Eighty-six percent of the 155 students completed a questionnaire based on the EU's DigComp 2.2 framework. Items (Likert scale and multiple-choice) assessed GenAI usage frequency and purposes, tool types, paid subscription use, critical evaluation capabilities, prompting proficiency, ethical concerns (privacy, bias, environmental impact, copyright), and perceived effects on academic behavior and career outlook. Results Daily use grew from 5% among first-years to 33% of Msc students. Paid subscriptions rose from 2.4% in first-year to 38% in second-year Master's students. ChatGPT was used by >90%. Educational use included brainstorming (56%−75%), summarizing (50%−76%), and text rewriting (44%−70%), coding assistance increased after the first year (16% vs. 44%−67% in later cohorts). Reliance on GenAI over traditional resources peaked in third-year undergraduates (45.8% agreement) vs. first year (9.1%). Competency ratings varied: Critical evaluation was high (60%−85% agreement). Prompting proficiency was moderate (35%−68%). Ethical appraisal was low (15%−56%). Competence in GenAI-assisted tasks (code writing, essay composition, practice question creation) was lowest among first-year (16%−38%) and highest in advanced cohorts (33%−83%). Privacy was the top ethical concern (46%−76%), followed by effects on skill development (38%−63%) and bias (33%−100%). Copyright (12%−46%) and environmental impact (13%−28%) received less concern. Interest in learning centered on prompting (58%−84%), privacy/security (60%−84%), and bias mitigation (38%−88%). Sixteen percent to thirty percent of students feared a negative impact of GenAI on their career perspectives. Conclusion Students increasingly use GenAI as a complementary academic resource but exhibit skill gaps in ethical safety practices and content creation. Institutions should ensure equitable, secure GenAI infrastructure and integrate ethics-centered AI literacy in the curriculum.