Using AI to understand health information safely
Artificial intelligence, often called AI, is now part of many websites, apps and search engines. Some people use AI tools to ask questions about symptoms, medicines, appointments, test results or caring responsibilities.
AI can be useful, but it has important limits. It should not replace advice from a doctor, nurse, pharmacist or other health and social care professionals.
What AI can help with
AI may help you:
- understand medical words in simpler language
- prepare questions for an appointment
- summarise information you already have
- organise your thoughts before speaking to a professional
- learn general information about palliative care, symptoms or support
For example, you could ask:
“What questions should I ask the nurse about managing pain?”
“Can you explain this medical term in plain English?”
“What information should I write down before phoning the GP?”
AI is safest when it is used to help you prepare, not to make decisions.
What AI should not be used for
AI tools are designed to be helpful. This means they may sometimes give answers that sound reassuring, confident or supportive, even when the answer is not fully correct.
AI can also appear to agree with the person asking the question. It may respond differently depending on how a question is worded.
For example, if you ask:
“Is this probably nothing to worry about?”
AI may give a reassuring answer.
If you ask:
“Could this be serious?”
AI may give a more worrying answer.
This does not mean either answer is correct. It means AI can be influenced by the wording of the question.
AI may also:
- sound certain when it is unsure
- give different answers to similar questions
- provide too much reassurance
- make a situation sound more serious than it is
- miss important warning signs
- give information that is too general for your situation
Because AI is designed to be helpful, it may not always challenge your assumptions or tell you clearly when you need advice from health and social care professionals.
Potential issues
AI can be eager to please
AI tools are designed to be helpful. This means they may sometimes give answers that sound reassuring, confident or supportive, even when the answer is not fully correct.
AI can also appear to agree with the person asking the question. It may respond differently depending on how a question is worded.
For example, if you ask:
“Is this probably nothing to worry about?”
AI may give a reassuring answer.
If you ask:
“Could this be serious?”
AI may give a more worrying answer.
This does not mean either answer is correct. It means AI can be influenced by the wording of the question.
AI may also:
- sound certain when it is unsure
- give different answers to similar questions
- provide too much reassurance
- make a situation sound more serious than it is
- miss important warning signs
- give information that is too general for your situation
Because AI is designed to be helpful, it may not always challenge your assumptions or tell you clearly when you need advice from health and social care professionals.
When to be extra careful
Do not use AI to decide what to do about new, worsening or worrying symptoms. If you are concerned, contact a health and social care professional, for example your GP, nurse or Consultant.
Do not use AI to decide whether to start, stop or change medicines. This includes prescribed medicines, over-the-counter medicines, herbal remedies and supplements. Speak to a pharmacist, GP, Consultant, nurse or palliative care team if you have questions about medicines.
If someone is seriously ill, injured or at risk of dying, call 999 (Northern Ireland and Republic of Ireland) or 112 (Republic of Ireland only).
Protecting your privacy
Think carefully before entering personal or sensitive information into an AI tool.
Avoid sharing:
- full names
- dates of birth
- addresses
- hospital or health and care numbers
- detailed medical letters or records
- names of family members or professionals
- photographs of yourself or someone else
Try to keep questions general. For example, instead of entering a full clinic letter, you could ask:
“What does ‘breathlessness at rest’ mean?”
or
“What questions could a carer ask about managing pain at home?”
Information for carers
If you care for someone with palliative care needs, AI may help you prepare for a conversation with a healthcare professional.
For example, you could use AI to help turn your notes into a clear summary before you contact a GP, nurse, pharmacist, out-of-hours service or palliative care team.
It may help you organise information such as:
- what has changed
- when the change started
- whether things are getting better, worse or staying the same
- what you have already tried
- what medicines the person has taken
- what you are most worried about
- what help or advice you need
This can make it easier to explain the situation clearly. AI should not be used to decide what care or treatment the person needs.
How to use AI more safely
If you choose to use AI to understand health information:
- treat it as general information only
- check important information with trusted sources or your healthcare team
- remember that AI can be wrong, even when it sounds confident
- be aware that AI may try to give a helpful-sounding answer, rather than the safest answer
- use it to prepare questions, not to make medical decisions
- contact a healthcare professional if symptoms are new, worsening or worrying
Remember
AI can help you understand general information and prepare for conversations.
However, AI can be wrong, overly reassuring or influenced by the way a question is asked.
It should not replace professional medical advice, diagnosis or treatment.
If you are worried about symptoms, medicines or changes in someone’s condition, contact a healthcare professional. If it is an emergency, call 999 (Northern Ireland and Republic of Ireland) or 112 (Republic of Ireland only).
Useful Links
Microsoft: Introduction to AI concepts
AbilityNet: What is AI and how do I use it?
Mayo Clinic: Can you trust AI for health advice?