A survey conducted by Harvard Medical School, published in JAMA Network Open, reveals that oncologists agree AI tools must be explainable, patients must consent to AI use, and oncologists must protect patients from AI biases. Despite this, many oncologists lack confidence in recognizing AI biases, highlighting a need for structured AI education and ethical guidelines. The survey found that 37% of oncologists would let patients decide between their own and AI treatment recommendations, and 77% believe they should protect patients from biased AI, though only 28% feel capable of identifying such biases.
Categories