\u201cIt [ChaGPT] can respond to questions, to respond to queries. However, it has a particular type of architecture, particular type of function. It actually gives a random output,\u201d said Dr. Michael Kosorok<\/a>, a W.R. Keenan junior distinguished professor of biostatistics and the director of the Center for Artificial Intelligence and Public Health at UNC Chapel Hill<\/a> said.<\/p>\n An example of how ChatGPT Health works from a demonstration with OpenAI. (OpenAI)<\/p>\n Kosorok, who is also a professor of statistics and operations research, said he is looking at finding ways to use artificial intelligence to help improve health care.<\/p>\n \u201cExamples include coming up with a easy to use, inexpensive, relatively inexpensive way of discerning gestational age in women who are in locations around the world that don’t have access to usual health care,\u201d Kosorok said. Another example of work he is conducting is developing algorithms based on data that can assign treatments to minimize chronic pain.<\/p>\n \u201cWe’re using artificial intelligence to solve various problems\u00a0that there are for helping patients in different situations across many different diseases,\u201d Kosorok said.<\/p>\n He said there are at least 20 projects across UNC-Chapel Hill\u2019s Gillings School of Global Public Health<\/a> that are trying to solve problems using artificial intelligence.<\/p>\n Kosorok said they use ChatGPT early in the process of creating their AI tool, but they remove it as they get closer to completion. His role is to test the system and make sure it’s reliable.<\/p>\n He said ChatGPT can help by showing scientists structure in the data that can be used in other models or tools.<\/p>\n \u201cOnce we get the tool working, it goes through various processes, and then it can be used by other physicians to apply to their situation or by other health care practitioners, or in some cases, to help making decisions about public health,\u201d Kosorok said. \u201cSome of the tools help the doctor to be able to make better decisions using data. It can inform their decision. The AI does not make a decision for the physician.\u201d<\/p>\n Other projects using AI at UNC<\/a> are in the works to help users find resources for health care challenges like opioid use. Other areas of developing projects are tools that could be useful for people who handle emergencies.\u00a0<\/p>\n \u201cFor us, the bigger the issues that we would look at is how good is the data for answering our questions? Does it have the information we need? And in many cases, we can collect data that’s extremely recent. And we always collect data that’s close enough and recency to be useful for the clinical,\u201d Kosorok said.<\/p>\n The University of North Carolina at Chapel Hill on Monday, April 13, will host the AI for Public Good Conference<\/a>. It will cover topics including AI for Precision Medicine, AI and the Arts and AI in National Security and Geopolitics.<\/p>\n The conference aims to bring together leading researchers, practitioners, policy-makers and others to learn how AI can be applied for the public good.\u00a0<\/p>\n Follow us on Instagram at\u00a0spectrumnews1nc<\/a>\u00a0for news and other happenings across North Carolina.<\/p>\n![\"An](\"https:\/\/www.europesays.com\/ai\/wp-content\/uploads\/2026\/04\/Screenshot_2026-04-10_at_7_08_53_AM.jpeg\")
<\/p>\n