The Limitations of AI Chatbots in Cancer Treatment Recommendations

A study published in JAMA Oncology found that OpenAI's ChatGPT, an artificial intelligence chatbot, scored poorly when providing cancer treatment recommendations. Researchers compiled 104 prompts and had four board-certified oncologists review the responses, resulting in an underwhelming score of 61.9%. The chatbot often provided unclear or mixed accurate and inaccurate information, with nearly 13% of responses being completely inaccurate or unrelated to the prompt. The study highlights concerns about misinformation and potential harm to patients. While AI chatbots show potential, they should be used as a supplement, not a replacement, for professional medical advice. Future research is needed to assess their long-term impact and generalizability in cancer treatment.