AI is a powerful tool, and like any tool, it needs to be used responsibly, especially in healthcare. This chapter covers key ethical points and good practices to keep in mind, ensuring you use AI safely and effectively while protecting the people you serve.

Keeping Patient Information Safe (Privacy and Confidentiality)

This is the MOST IMPORTANT rule: NEVER put personal, identifiable patient information into public AI chatbots like ChatGPT, Copilot, or Gemini.

• What is identifiable information? This includes:

  • Patient's full name, national ID number, address, phone number.

  • Specific details about their medical history that could easily identify them, even without a name.

  • Photographs or videos of patients.

• Why is this so important?

  • Protecting Patients: Health information is private and sensitive. Sharing it without permission is a breach of trust and can harm the patient. Zimbabwe has a Data Protection Act that you must follow.

  • AI Tool Policies: When you use free public AI tools, the information you type in might be used by the AI companies to improve their systems. It's not a secure place for confidential patient data. (Nature: Ethics of ChatGPT in medicine - highlights privacy issues)

• How to use AI safely for case-related queries:

  • Ask general questions: Instead of "Amai Moyo in village X has [symptoms], what could it be?", ask "What are common causes of [symptom type] in adults?"

  • Use anonymised scenarios: If you need to brainstorm ideas for a type of situation, describe it in general terms without any details that could identify a real person. For example, "How can I explain medication adherence to an elderly patient who often forgets?" rather than "Mr. Dube at [address] forgets his TB meds..."

AI Can Make Mistakes (Accuracy and Verification)

AI tools are very clever, but they are not perfect. They can sometimes:

• Give incorrect information.

• "Hallucinate" – this means they make up facts or details that sound believable but are not true (PMC: ChatGPT for LMICs - mentions hallucinations).

• Have outdated information, as their knowledge is based on when they were last trained (though some, like Copilot, can search the live internet).

CRITICAL ACTION: Always double-check important health information you get from AI.

• Verify with official Ministry of Health and Child Care (MoHCC) guidelines.

• Check with your supervisor or a qualified clinical colleague.

• Consult trusted medical textbooks or resources.

AI is for assistance and idea generation, not for making final decisions about a person's health, diagnosis, or treatment. Your professional knowledge and official protocols must always come first.

Being Fair and Avoiding Bias

AI learns from the vast amount of information created by humans. Sometimes, this information can contain human biases related to gender, race, culture, or economic status. This means that AI responses might occasionally reflect these biases, even if unintentionally (Nature: Ethics of ChatGPT in medicine - mentions bias).

• Be a critical thinker: When you get an answer from AI, think about whether it seems fair and appropriate for everyone in your community.

• Consider different perspectives: If AI gives you one idea, think if there are other ways to approach the situation that might be more inclusive or suitable for your specific local context in Zimbabwe.

• If an AI response seems biased or unfair, don't use it. Try rephrasing your question or seeking information from other sources.

Getting Permission (Informed Consent in Community Interactions)

While you won't be inputting patient data, if you are using AI to help you create educational materials that you will then share with community members, or if discussing a general (anonymised) scenario inspired by community needs:

• Always ensure your interactions with community members are respectful and patient-centered.

• When sharing information (even if AI helped you draft it), ensure it is accurate, easy to understand, and appropriate for the audience.

• The focus should always be on empowering the community member with knowledge, not just on using a new tool.

Your professional conduct remains paramount. This is more about your interaction with people after using AI, rather than inputting their data, which you must not do.

You Are Still the Professional! (Maintaining Professional Judgment)

This is a crucial reminder that we've mentioned before, and it's worth saying again:

AI is a tool. You are the skilled, compassionate Social Health Worker.

• Your local knowledge, your understanding of your community's culture and needs, your ability to build trust, your empathy, and your professional judgment are irreplaceable.

• Use AI to support your work, to make tedious tasks faster, or to get new ideas. But never let it replace your critical thinking or your direct, caring interaction with the people you serve.

• You are in control of how you use the information or drafts AI provides. You decide what is appropriate, accurate, and helpful for your community in Zimbabwe.

Think and Do: True or False? AI Ethics Quiz

Read each statement and decide if it's True or False.

1. It's okay to type a patient's full name and village into ChatGPT if I want to get advice on their condition. (True/False)

2. Information from AI chatbots is always 100% correct and up-to-date. (True/False)

3. I should always double-check important health information from AI with official MoHCC guidelines or my supervisor. (True/False)

4. AI tools cannot replace my understanding of my local community and culture. (True/False)

5. If AI helps me draft a health message, I don't need to review it before sharing it with the community. (True/False)

Answers: 1. False, 2. False, 3. True, 4. True, 5. False