The growing adoption of artificial intelligence is expected to make a major impact on the healthcare sector in every area, from physician education to treating patients, although understanding how quickly that adoption is progressing compared to other industries has been somewhat of an open question due to a lack of data.
A group of researchers, including Distinguished Professor Kosali Simon from the Paul H. O’Neill School of Public and Environmental Affairs, is attempting a new approach to fill the knowledge gap for the first time.
In the article, “Adoption of Artificial Intelligence in the Health Care Sector,” published in the JAMA Health Forum, Simon and colleagues from the University of Michigan, Brown University, Harvard, and the RAND Corporation analyzed the United States Census Bureau’s Business Trends and Outlook Survey to examine changes in AI use from September 2023 to May 2025. The study found that although the use of AI by healthcare firms increased over time, its use was lower than in other sectors, such as finance and insurance, education, and information services.
“Our census-scale data allow healthcare to be viewed alongside every other sector, revealing where adoption patterns are unique and where they reflect broader economic trends,” Simon said. “Many previous studies of AI in healthcare fielded surveys only of that one sector and, thus, the comparisons with economy-wide trends are new to the literature.”
The BTSO survey covers more than one million firms across the entire economy.
The largest gains in the healthcare sector were found for outpatient and ambulatory care, as well as nursing and residential care facilities.
“The healthcare workforce faces strain from documentation, regulatory requirements, and operational complexity,” Simon said. “AI has the potential to reduce administrative load, strengthen patient-centered care, and support discovery and clinical decision making. By comparing healthcare with other sectors, we can identify where adoption lags, where it leads, and enable next stages of research on which bottlenecks might be addressed to allow AI to support clinicians, patients, and health systems more effectively.”
The next step in the research will include a shift from analyzing data to real-world questions and AI’s impact on patient outcomes, costs, quality, and safety.
“Understanding the consequences of adoption will help guide decisions about where AI can deliver the greatest value for the healthcare sector,” Simon said.