People around the world are increasingly turning to ChatGPT for help with health-related questions. According to a report by OpenAI that analyzed anonymized messages, healthcare-related topics account for more than 5% of all messages on ChatGPT. That means billions of messages per week. Of more than 800 million regular users, one in four asks about health matters every week, and more than 40 million people do so daily. In the US, where the healthcare system is often complex and expensive, three in five adults say they have used AI tools for their health or healthcare in the past three months.
Query Analysis
A survey conducted by OpenAI using the Knit tool among 1,042 American adults who had used AI for healthcare in the past three months found that 55% of people had checked or researched symptoms. Another 48% wanted to understand medical terms or instructions, and 44% asked about treatment options. Users send 1.6 to 1.9 million messages about health insurance each week, asking about plan comparisons, prices, claims processing, billing, eligibility, and cost-sharing details.
In underserved rural areas of the US, home to about one-fifth of the population, users send an average of nearly 600,000 healthcare-related messages per week. OpenAI defined "hospital deserts" as places more than a 30-minute drive from the nearest general or children's hospital. For example, in such areas in Oregon, Texas, or New York, people often seek information when care is far away. Healthcare professionals also use AI: according to an American Medical Association (AMA) survey, 66% of US physicians used AI for at least one purpose in 2024, compared with 38% in 2023. Among nurses, 46% used it weekly, according to a survey by Wolters Kluwer.
Real-World Examples
For example, San Francisco tax specialist Ayrin Santoso used ChatGPT to help coordinate urgent care for her mother in Indonesia, who had experienced sudden vision loss. After entering the symptoms and context, she received a warning about a possible hypertensive crisis and stroke, which led to an immediate blood pressure check and hospitalization. Another case involves Rich Kaplan from Seattle, who has a rare autoimmune blood-clotting disorder. Using ChatGPT, he found studies and prepared an appeal against his insurer's denial of therapy, which ultimately succeeded. In Montana, in the small town of Miles City with 8,400 residents, family physician Dr. Margie Albers uses tools based on OpenAI models, such as Oracle Clinical Assist, to write visit notes and reduce administrative work, allowing her to focus on patients who travel for hours.
OpenAI emphasizes that AI helps in areas such as repurposing approved drugs. For example, immunologist Dr. Oral Alpan of McLean, Virginia, used GPT-5 Pro to test whether the drug dupilumab helps patients with FPIES, a rare food allergy. Based on a de-identified case description, the model correctly suggested the drug. OpenAI plans to release policy recommendations soon, including better integration of medical data to accelerate discoveries.
Additional sources: axios.com and independent.co.uk
