Understanding and Consent: 85% of oncologists believe they should understand how AI models function, yet only 23% think patients need the same depth of understanding. However, 81% agree that patient consent is essential when AI aids treatment decisions.

Decision-Making: When AI suggests a different treatment than the oncologist, 37% would offer both choices to the patient, highlighting the value of shared decision-making.

Responsibility: A significant 91% of oncologists think AI developers should be accountable for medical or legal issues arising from AI use, far exceeding those who believe responsibility should also involve physicians (47%) or hospitals (43%).

Bias Protection: While 76% of respondents recognize the need to shield patients from biased AI tools, only 28% feel confident in their ability to detect such biases.

The study underscores the importance of involving oncologists in discussions about AI's ethical implications and future roles in cancer care, especially as AI's applications, such as in diagnosis and potentially treatment recommendations, continue to evolve. This research initiates critical conversations on how AI should be ethically integrated into cancer treatment, ensuring it aligns with patient care standards and legal responsibilities.