ChatGPT is functioning as a de facto health insurance helpline. OpenAI's Head of Business Finance, Chengpeng Mou, released anonymized U.S. usage data showing roughly 2 million weekly messages about health insurance alone.
The access gap is the real story. Approximately 600,000 of those weekly messages come from users in hospital deserts, defined as locations more than a 30-minute drive from the nearest hospital. Seventy percent of all healthcare-related messages are sent outside clinic hours.
These are not vanity metrics. They are behavioral data showing where the American healthcare system fails people and where they are turning instead. The full data raises urgent questions about liability, accuracy, and what it means when a language model becomes primary care infrastructure.
[READ ORIGINAL →]