Using AI for health advice? – ask SMART

A practical guide on using AI chatbots to ask health questions

Download guide

– ask SMART

AI chatbots can be useful for health questions, but they can also get things wrong. – ask SMART

They might sound like a person, but they work like a machine

This guide is for anyone who might use an online AI chatbot to ask health questions. ChatGPT, Gemini, and Copilot are examples of AI chatbots. They can be used for many different things.

Most AI chatbots were not designed specifically for health, but many people find them helpful for health questions. They are not perfect though, and they can get things wrong.

Some ways of using chatbots might be more risky than others – for example, using them during a mental health crisis. It's up to you to decide if, and how, you want to use them.

This guide will help you get the most out of AI chatbots for your health. You'll learn how to write clear questions, think carefully about the information you share, check the facts, and know when a problem is best solved by a health professional.

Click the other tabs to learn more.

Find out more

AI chatbots cannot read your mind

Your questions or messages to a chatbot are called 'prompts'. The way you write prompts will change the answers you get. Here are some things you could try:

• Tell the chatbot how you would like your answers; long, short, simple language, bullet point list etc.

• Tell the chatbot to ask you questions. Your answers might help it produce more helpful responses.

• If you get a response that isn’t clear or doesn’t answer your question, ask follow-up questions, or start a new conversation and rephrase the prompt.

You can change how AI chatbots work by changing their settings. This lets you set rules for how you want the chatbot to respond, and how it looks and feels to use. Most also have advanced features, like ‘skills’ and tools for deeper research.

Find out more

Once it’s sent, it’s out of your hands

Be careful about sharing personal or sensitive data with AI chatbots. Things you share with an AI chatbot might not stay private. This could be because of hacking, data leaks, or the chatbot using or sharing your data.

AI chatbots may not have access to your medical history or information about your lifestyle. This means their responses may be generic, and less suited to you. If you share personal information, the amount of detail you give can affect how useful the response is. Sharing more may help the chatbot respond to your specific problem, but sharing less may help protect your privacy. Only share information that relates to your question, and remove names, addresses and other personal details from anything you copy and paste or upload.

Avoid sharing private details like identity numbers or your full address with AI chatbots. If this gets into the wrong hands, it could be used to steal your identity.

Red flag:

Only share information that’s relevant to the question you’re asking.

Find out more

AI chatbots can be wrong and still sound confident

AI chatbots can say things that are not true. They can also show bias or discrimination, and agree with you even when you are wrong. This could persuade you to choose treatment options which cause you harm.

Some AI chatbots can list sources to back up the answers they give. This can be reassuring, but the sources don’t always match the chatbot’s answers.

Before making important health choices, check the answers given by AI chatbots against independent, trusted sources of health information. These include official health services, medical charities, international bodies like the World Health Organization, and qualified healthcare professionals.

Red flag:

AI chatbots can get things wrong. Check the facts before making important decisions that affect your health. If you are worried, do not delay getting help because a chatbot says your symptoms are probably harmless.

Find out more

There are some things that AI chatbots just cannot do

Do not use an AI chatbot in an emergency. Seek immediate medical help instead.

Talk to a qualified healthcare professional or someone else you trust if:

• You think a chatbot is giving you incorrect information,

• The conversation is going around in circles,

• You find its responses confusing,

• Your condition is getting worse,

• You are thinking about starting or stopping a medication,

• Or you are not sure what to do.

Using an AI chatbot is your decision. You don't need to tell a healthcare professional you've used them if you don't want to. But if its response has worried you, it may be helpful to talk to someone about it.

Red flag:

If the conversation with an AI chatbot starts to feel distressing, inappropriate, unusually important, or simply isn't going anywhere, stop and speak to someone you trust.

Find out more

Supported by

The Health Foundation
National Institute for Health and Care Research (NIHR) Biomedical Research Centre: Birmingham
University of Birmingham

What is the guide all about?

This short video introduces the Health Chatbot User Guide, and explains how you can Ask SMART

Trusted by health professionals and the public

Health chatbots can be helpful when you’re trying to make sense of health information. But just because an answer sounds convincing doesn’t mean it’s right. A fantastic global team has put this guide together drawing on conversations with people who use these tools.

Dr Deborah Cohen

Dr Deborah Cohen

Steering group member and author of Bad Influence: How the internet hijacked our health

As a Public Contributor, I appreciated the welcome to share in the creation of this guide. I am confident that reading it and following its advice will enable the public to take charge of their relationships with AI chatbots.

Rosemary Nicholls

Rosemary Nicholls

Public contributor

The changing landscape of AI and its widespread use by the public calls for a project like this. It is a much-needed initiative to spread evidence-based science to a wider audience.

Dr Lorenzo Righetto

Dr Lorenzo Righetto

Senior editor at Nature Health

At last, a plain English guide to using AI chatbots when searching for health information. Read it before you ask your next health question.

Moira Auchterlonie

Moira Auchterlonie

Public contributor

Why use this guide

Who wrote it?

The Health Chatbot User Guide is a collaborative project led by Bright at the University of Birmingham. Our steering group included members of the public, AI experts and health professionals from around the world.

We’re independent

This project was supported by The Health Foundation and the National Institute for Health and Care Research Birmingham Biomedical Research Centre. We did not receive any funding from AI or technology companies.

Plain language, no jargon

No complex technical terms, just clear, practical guidance in language anyone can act on

FAQs

It's a free guide that helps you get more out of AI chatbots when you're asking about your health - and know when to stop and speak to a real person instead. It's built around five simple steps: Ask SMART.

No, we're not here to tell you what to do. We'll help you get more out of them and spot when something's higher risk, so you can make your own call.

We are a team of health researchers at the University of Birmingham. We're not owned by, or paid by, any chatbot company. The guide was co-developed with members of the public, AI experts and health professionals around the world.

No. The guide isn't tied to any particular chatbot. The 5 steps we cover work whichever one you're already using.

Yes and yes. The Guide is free, and you can read it without creating an account or giving us any personal details.

AI chatbots change fast, so we will review the Guide and update it in the future.

If you think you or someone close to you is having a health emergency or is in immediate danger, get help from a trained health professional straight away. If you are in the UK, call 999 or go to A&E as soon as possible.

Yes, the guide has advice about using a chatbot to ask about someone else's health. Remember, you may need their permission before you share anything personal or sensitive.

AI chatbots are tools which can help you manage your health, but they cannot replace healthcare professionals. There are some things that AI chatbots can't do, so it's important to speak to a trained health professional too.

A consensus-based guide to using AI chatbots for health questions

By

Bright

Sign up to our newsletter

Keep up to date with the latest developments of The Health Chatbot User Guide

By signing up, I confirm that I’m over 16. To find out what personal data we collect and how we use it, please visit our Privacy policy

© 2026 The Health Chatbot User Guide authors, CC BY 4.0. Birmingham, United Kingdom University of Birmingham; NIHR Birmingham Biomedical Research Centre

Designed with

by Bridge Studio