Large language models such as GPT-4 have emerged as powerful new tools for information work, including aspects of healthcare delivery, healthcare management, and advancing medical science. GPT-4 in particular shows impressive capabilities that go beyond direct medical knowledge, to include even support for patient engagement and medical research. This talk presents highlights of an intensive, year-long study exploring the benefits, limitations, and risks of GPT-4 in medicine. Our fundamental conclusion is that GPT-4 may represent the most important advance in computing technology, with potentially transformative impact on healthcare, and despite having received no specialized training in medicine. Examples will be shown of how the general intelligence of GPT-4 can be used, with implications for the current and future practice of medicine.

Learn more about the AI Forum 2023 hosted by Microsoft Research Asia in collaboration with The University of Tokyo: https://www.microsoft.com/en-us/research/event/ai-forum-2023/overview/