2436 - Advanced Practice Providers vs. Attending Physicians: Differential Readiness and Attitudes Toward AI Autonomy in Radiation Oncology

Purpose/Objective(s): Advanced Practice Providers (APPs) are integral to radiation oncology care teams. This analysis compared APP and attending physician (ATT) perspectives on AI autonomy levels, ethical boundaries, and adoption readiness.

Materials/Methods: Subgroup analysis of 15 radiation oncology clinicians with defined roles (5 APPs, 10 ATT) from a January 2026 anonymous survey at a comprehensive cancer center was performed. Providers rated comfort (on 5-point Likert scales) with increasing AI autonomy levels (L1 – basic support; L2 – routine task handling; L3 – collaboration suggests options and flags concerns; L4 – clinical guide recommends treatments but requires approval; L5: fully autonomous practice, without clinician approval), supported by RO-specific clinical and technical vignettes. L5 examples included independent management of toxicities or independent adaptive treatment planning/delivery. Multiple domains of adoption were assessed. Descriptive statistics and Friedman test were used.

Results: Baseline daily or weekly personal AI use differed significantly APP 0% vs ATT 50%, while clinical use was ~40% for both. Comfort with increasing AI autonomy declined similarly across groups and was low at full autonomy (L4: both 40%; L5: APPs 20% [mean 2.0] vs ATT 10% [mean 1.7];). All APP and 90% of ATT felt human clinical judgment will remain essential for care. APPs demonstrated markedly higher readiness for validated collaborative AI adoption (L3-L4, 80% within 2 years vs 40% of ATT) and were more likely to project =51% workflow replacement (40% vs 20%). Both groups expected moderate burnout reduction from agentic AI (60%) and viewed hospitals/health systems as the leading adoption driver (60% APPs, 30% attendings). Personal experimentation was the top influencer for use for APPs (80%) vs ATT (30%), compared with published evidence for ATT (50%) vs APP (20%).

Conclusion: Radiation Oncology APPs and attendings share strong core ethical boundaries for implementation of AI in an oncology care framework and are comparably supportive for early use AI, with mutual reluctance for autonomous L5 workflows. They share similar views on accountability and consent and have general optimism regarding AI- driven burnout reduction. RO APPs anticipate much faster integration of agentic AI than attending physicians, despite having lower baseline exposure to current tools. The higher comfort reported by RO APPs toward Level 4 "guide" workflows may reflect their existing structural familiarity with collaborative, sign-off-based clinical practice. Additionally, the RO APPs anticipation for faster AI adoption and their increased likelihood to adopt AI systems based on personal experimentation, may suggest that APP-led implementation pathways may accelerate practice integration if aligned with evidence-based guardrails.