59-year-old Burt Rosen is not your average mountaineer and marketing guy. He also happens to be battling two different types of cancer—a kidney cancer and metastatic pancreatic neuroendocrine tumors. Managing both of these complex cases requires a whole network of specialists. Rosen’s patient portals are quite overwhelming, which led him to turn to AI for help. He recently asked ChatGPT to translate his CT scan results into simpler language that he could understand. Within seconds, the AI provided him with a summary in bold subtitles, making it easier for him to grasp the information.
Although many people have yet to dive into the world of artificial intelligence, it is already making waves in the healthcare industry. Doctors are utilizing AI to assist in diagnosing and documenting, while patients like Rosen are leveraging its capabilities. According to a poll by KFF, one in six Americans already use AI monthly for health information and advice. But is this widespread use of AI a good idea?
Chatbots like ChatGPT, Claude, Perplexity, Copilot, and Gemini are considered immensely powerful AI models known as large language models (LLMs). They can be both incredibly smart and surprisingly naive, as computer scientist Yejin Choi pointed out. While ChatGPT achieved impressive scores on medical exams, it is not immune to errors. Sometimes, it can provide incorrect or fictitious information, known as “hallucinations.”
Unlike conventional software programs that are coded for specific tasks, LLMs like ChatGPT gather information on their own, making it unclear what goes on inside them. This lack of transparency means that troubleshooting and reprogramming them is not feasible. Therefore, AI will not be replacing primary care physicians anytime soon. However, chatbots can be valuable tools in navigating one’s health journey if used correctly and cautiously.
Isaac “Zak” Kohane, M.D., Ph.D., editor-in-chief at NEJM AI, emphasizes the importance of knowing how to harness the power of AI and avoid potential pitfalls. AI can help individuals understand their health conditions better, decode medical jargon, provide second opinions, and suggest alternative diagnoses. It can offer a holistic view of one’s health, which may be beneficial when traditional specialists have failed to provide answers.
For instance, Courtney Morales Hofmann’s son was undiagnosed for three years until AI suggested a likely cause based on his medical records. While AI can be a valuable resource, it is essential to approach it with caution. Some studies have shown that chatbots may provide inaccurate information, so it is crucial to verify sources and cross-check the advice given.
AI can also assist individuals in getting detailed information, exploring potential diagnoses, seeking second opinions, and even acting as an ally in legal matters. It can help users delve deeper into their health data, providing insights and trends that may not be apparent otherwise. Wearables that track health metrics can use AI to process data and offer personalized recommendations for better health management.
By being informed and empowered, individuals can become effective partners with their doctors. AI can help in various aspects of healthcare, from creating visit summaries and patient education materials to streamlining literature searches and providing alerts for potential health issues. While AI may not replace human doctors entirely, it is already playing a significant role in enhancing healthcare delivery and patient outcomes.
