OpenAI rolls out ChatGPT Health to all US users and touts its accuracy

In short: OpenAI is expanding ChatGPT Health to everyone in the US, letting users connect medical records and health tracking data to ChatGPT.

What happened

OpenAI says ChatGPT Health is rolling out to all US users on Thursday. The feature lets people link medical records and health tracking information to the chatbot, then ask questions based on that information.

OpenAI health product leader Ashley Alexander said in a briefing that the company’s AI models can now reason “better than clinician level.” A clinician is a trained medical professional, like a doctor or nurse.

When asked to explain the claim, OpenAI health lead Karan Singhal said it should be “tempered.” He added that some individual studies suggest AI can outperform doctors in certain diagnosis tasks, including a study from Harvard and Stanford that was referenced in the briefing.

The update also changes where the feature lives inside ChatGPT. Instead of needing a separate ChatGPT Health tab, users can ask health questions in the main chat. OpenAI says ChatGPT will ask permission before using connected health data to personalize an answer, and that conversations are encrypted (locked with a digital code while stored and while being sent).

The rollout comes as concerns about medical advice from chatbots continue. The Verge notes that a Florida pastor sued OpenAI this week, claiming ChatGPT gave “extremely dangerous medical recommendations” and that he delayed care for a pulmonary embolism.

Why it matters

Linking medical records to a chatbot could make it easier to understand lab results, medications, and visit notes, like having a helper that can read a messy folder of paperwork. But OpenAI’s bold performance claims, plus recent legal complaints about unsafe advice, highlight the risk if people treat chatbot answers like a real diagnosis.

Source: The Verge AI