Chronic illness patients are increasingly turning to AI chatbots like Claude for diagnoses after failing to find answers through traditional healthcare systems. A New York Times investigation reveals women with complex conditions such as long COVID and dysautonomia—where bodily functions like heart rate and digestion become unpredictable—are using these tools to piece together their symptoms. One patient, Margie Smith, 70, of North Carolina, credits Claude with identifying her conditions after years of misdiagnoses across multiple specialists. “The medical system really failed me,” she said, though she acknowledges AI’s limitations.

The trend reflects broader frustrations with fragmented care. A March poll found 33% of adults use health chatbots, including those with poorly understood illnesses like autoimmune diseases. While some, like Caroline Gamwell, a pelvic floor therapist who diagnosed herself with fibromyalgia via ChatGPT, praise AI’s ability to synthesize data, experts warn of risks. Dr. James Landay of Stanford’s AI institute notes chatbots can hallucinate or pull from unreliable sources, leading to dangerous decisions. A February study showed non-experts using chatbots correctly diagnosed cases less than half the time.

Despite pushback from health institutions, patients like Patty Costello—who discovered mast cell activation syndrome through ChatGPT—highlight AI’s potential. Costello, who saw 10 diagnoses from ChatGPT and pursued an allergist, now feels 80% improved. Yet Dr. John Whyte of the American Medical Association cautions, “This is a new version of an old story” of self-diagnosis, amplified by AI’s persuasive authority. OpenAI and Anthropic declined to comment, though OpenAI emphasized chatbots “are not a substitute for professional advice.”

The reliance underscores systemic gaps: women’s symptoms are often dismissed, and diagnoses can take years. While AI remains controversial, its role in bridging healthcare gaps seems inevitable. As one patient put it, “It’s the option that’s available.”