OpenAI launches ChatGPT Health, can the chatbot serve as a personal health assistant?

OpenAI today announced the launch of “ChatGPT Health,” allowing users to connect their personal medical records and health applications within ChatGPT to create a personalized consultation space. This is seen as an important step for OpenAI to further expand its “personal super assistant” positioning.

What is ChatGPT Health? Your personal health assistant is now online

As large language models (LLM) become integrated into daily life, health and medical inquiries have become one of the most frequently consulted fields by users. OpenAI states that ChatGPT Health, like the “Library (Library)”, is an independent category within ChatGPT, dedicated to handling conversations related to health, medical care, and lifestyle nutrition.

This feature is not intended for disease diagnosis or treatment but aims to assist users in understanding information, organizing health records, and providing responses that are more tailored to individual contexts in daily life.

The company emphasizes that files, conversations, and linked data within the Health section are stored separately from general chats to avoid affecting other usage scenarios.

From Apple Health to Medical Records: Multi-Data Integration for Recommendations

ChatGPT Health allows users to independently connect various health and lifestyle records, including data from Apple Health, nutritional advice from MyFitnessPal, personalized diet guidelines from Weight Watchers, and blood test data from Function, among others:

You can ask, “Based on my medical records over the past few years, help me see which insurance plan suits me better,” or “I’m taking a certain medication but also want to rebuild muscle mass. Can you help me plan a weekly meal schedule?”

OpenAI has also partnered with health tech company b.well to enable users to upload personal medical records and other health data into the Health section for more accurate responses.

Privacy and Model Training Segregation, OpenAI: Health Data Will Not Be Used to Train AI

Regarding privacy and data usage, OpenAI reiterates that conversations within ChatGPT Health will not be used to train its foundational models, and related information will not flow into other chat scenarios: “All data links are decided by the user and can be interrupted at any time.”

Fidji Simo, CEO of OpenAI’s application division, pointed out that this design is to address users’ concerns about sensitive health data and to ensure that the Health section is a closed and controllable environment.

(Will AI Replace Medical Personnel in Prescribing Medications? Utah, USA, Launches Pilot Program)

Demand and Risks Coexist: The Reality of AI Intervention in Healthcare

OpenAI revealed that over 230 million users ask health and medical questions on ChatGPT weekly, indicating a huge demand. However, external experts also remind that the operation of LLMs is based on probability prediction rather than judgment of “accuracy,” and there are still risks of hallucinations and misinformation.

OpenAI explicitly states in its terms of service that ChatGPT Health is not suitable for diagnosing or treating any health conditions. It will be initially opened for testing to a small group of users in the coming weeks and gradually expanded.

This article about OpenAI launching ChatGPT Health—can a chatbot serve as a personal health assistant? first appeared on Chain News ABMedia.