Background
Inpatient based oncology education for junior doctors and medical students is often inconsistent and limited because of competing clinical workloads. There is an increasing need to modernise medical training and recent evidence supports the role of artificial intelligence (AI) and large language models (LLMs) in enhancing clinical reasoning and medical learning outcomes. We aimed to investigate whether AI could be integrated into regular bedside teaching in a regional Australian cancer centre with benefit to our students.

Methods
A pilot study was conducted over eight weeks. Teaching was delivered to junior doctors and medical students daily during weekday ward rounds in 20-minute micro sessions. Four oncological themes were covered: foundational oncology, systemic therapy, patient-centred care, and tumour-specific learning.
Each ‘Smart Round 2’ followed a four-step format:
1. Pre-session MCQs (generated using generative pre-training transformer (GPT))
2. AI-generated clinical scenario
3. Facilitated group discussion
4. Post-session reflection questionnaires and repeat MCQ

Results
Twenty-five junior medical staff participated voluntarily: 10 Year 4–5 medical students, 9 resident medical officers (RMOs), and 6 basic physician trainees (BPTs)
•100% of sessions ran as planned
•95% found sessions clinically relevant and efficient
•88% reported improved understanding of oncology topics
•92% supported program continuation
•21% improvement in post-session mean MCQ scores

Emerging themes:
•Authenticity: AI-generated cases reflected real-world complexity
•Clarity: Explanations were concise and clinically aligned
•Engagement: Repetition and discussion improved retention
•Flexibility: Curriculum adapted to learner input (e.g., pain management, steroid tapering)

Conclusion
The integration of AI and LLM assisted micro teaching sessions is feasible in a regional Australian cancer centre. Unlike passive e-learning or simulation-only approaches, this model delivers structured, real-time instruction using AI agents in authentic clinical settings. The intervention was perceived highly positively amongst students and junior doctors. An improvement in knowledge was demonstrated. Further research in this area is highly desirable.