Artificial intelligence is transforming the way people search for health information. Tools like ChatGPT Health, a new feature from OpenAI developed with input from medical professionals, answer questions about symptoms, conditions and wellness more conversationally than ever before. But while these capabilities make health knowledge more accessible, they come with important limitations. Understanding what AI can — and cannot — do is critical to using it safely.

What ChatGPT Health Can Help With

ChatGPT Health can be a useful companion for general health questions when you use it within its intended scope:

1. Clarifying Medical Terminology: It can explain clinical terms, break down what lab tests mean and outline how common conditions develop — information that may make conversations with your doctor easier to navigate.

2. Exploring Symptoms and Conditions: Asking broad questions like “What are typical symptoms of seasonal flu?” or “How does hypertension affect daily life?” can yield well-organized summaries rooted in established medical knowledge.

3. Preparing for a Consultation: AI can help you think through questions to ask your clinician, highlight topics worth raising, and outline general lifestyle or prevention steps that align with accepted medical guidance.

4. General Health Education: Topics such as vaccine benefits, nutrition basics and safe fitness habits are well suited to AI explanation and can support health literacy.

These uses make AI a practical resource for foundational understanding and planning — but not a substitute for professional diagnosis or care.

Clear Limits: What ChatGPT Health Is Not Designed To Do

Despite its capabilities, ChatGPT Health has built-in restrictions and important gaps:

1. No Personal Diagnosis: It cannot assess your unique medical history, physical exam findings, imaging, or lab results in context. Without this, accurate diagnosis simply isn’t possible.

2. No Treatment Plans or Prescriptions: What works for one person may be unsafe for another. AI lacks the clinical judgement and legal accountability required to recommend medications, dosages or tailored treatment strategies.

3. No Replacement for Clinical Judgement: Health decisions often hinge on nuanced judgement, ethical considerations, and legal standards that AI cannot replicate.

4. Risk of Inaccurate Responses: Even well-written answers may be outdated, incomplete or misleading — a known risk in AI outputs.

5. Policy Restrictions: OpenAI’s own guidelines prevent the model from offering individualised medical advice that would normally require licensed practice.

For these reasons, AI should support, not replace, human clinical care.