Artificial intelligence and clinical reasoning - a qualitative study with general practice trainees in Germany

Background General practitioners face high complexity in clinical decision-making. Large language models (LLMs) may support clinical reasoning, but their impact on decision-making processes and postgraduate medical training remains unclear. This qualitative study aims to explore General Practice (GP) trainee’s use of LLMs in clinical reasoning and their educational needs. Materials and methods GP trainees were recruited and interviewed with semi-structured interview-guide. Interviews were transcribed verbatim and analysed qualitatively. Results Seventeen GP trainees participated (mean interview duration: 38 min; range: 21–63 min). AI was perceived as a tool to augment and accelerate clinical decision-making, while potential risks of deskilling were identified. Participants reported a lack of structured AI education and expressed a need for its integration into postgraduate training. Discussion Participants perceived AI primarily as a supportive tool that can complement clinical reasoning and improve access to medical knowledge. At the same time, they identified potential risks, including overreliance on AI-generated information and possible effects on the development of clinical expertise. These findings suggest that postgraduate medical training should address AI-related competencies, including critical appraisal, ethical responsibility and appropriate integration of AI into clinical decision-making. Conclusion AI is already part of clinical practice among GP trainees, offering opportunities to support learning and clinical reasoning while raising concerns about overreliance and critical appraisal. Postgraduate medical training in GP should therefore incorporate structured AI competencies to ensure responsible, reflective and patient-centered use of AI.