Using AI well involves more than getting a useful answer. It also means making thoughtful choices about the information you provide and the people your output may affect. Part four of our beginner series gives you three places to start.
First, consider the information you share
A prompt can contain someone’s personal information just as a document can. New Zealand’s Office of the Privacy Commissioner recommends understanding how AI tools handle information and advises against using them to handle personal information when you are in doubt.
Further reading: New Zealand Privacy Commissioner’s AI guidance
For practice, write a fictional example from scratch. If you want help with the structure of a customer reply, invent a customer and a situation instead of pasting a real complaint. You can often learn the writing technique without using the real record.
Before using work material, check which tools and data your organisation permits. Review the relevant privacy settings and terms rather than assuming that every account handles information in the same way. Removing a name alone may not remove identifying details.
Next, look for the perspective that is missing
AI outputs can reflect harmful biases and stereotypes. NIST identifies these as risks to consider. Treating an output as neutral simply because software produced it can hide assumptions that deserve attention.
Background: NIST’s discussion of harmful bias in generative AI
Suppose a tool drafts a community event announcement that assumes everyone drives, has reliable internet, and can attend during working hours. Instead of accepting those assumptions, check the announcement against the actual audience and the options the event really offers.
This prompt helps you make a review list. It does not prove that the final text is inclusive. When a message concerns a particular community, invite relevant people to review it rather than asking a model to speak for them.
Keep responsibility with people
UNESCO’s student AI framework emphasises human agency and accountability alongside technical understanding. In everyday use, that means deciding whether a tool is appropriate and taking responsibility for the material you choose to share.
Further reading: UNESCO’s AI competency framework
Before sending an AI-assisted message, ask: would I stand behind these words? Have I checked the facts? Does my workplace, course, or audience expect me to explain how AI was used? If you cannot answer those questions, pause and resolve the uncertainty.
Make one practical change today
Create a small fictional practice scenario that you can reuse. Write down what information is unnecessary for the exercise and which assumptions you want to check. You will have a useful learning task without needing to share a real person’s story.

