What does responsible AI use look like for you?

Most of the conversation about AI ethics happens at the level of governments and large companies. That matters, but it can leave ordinary users wondering what any of it means at their own desk. The good news is that the same principles scale down. Fairness, transparency, privacy and accountability all have a personal version, and each one can be turned into a question you ask yourself in a few seconds.

A Beginner's Guide to ChatGPT makes the starting point plain: these tools can be wrong, and they can invent details that sound real, so a person has to review what comes out. Responsible use builds on that. It is less about grand rules and more about habits that keep you, and the people affected by your work, out of trouble.

Four principles, made personal

In November 2021, all 193 member states of UNESCO, the United Nations agency for education, science and culture, adopted the first global Recommendation on the Ethics of Artificial Intelligence. It is written for governments, but its main themes translate directly into daily choices.

The Recommendation also asks for methods that use less data, energy and resources. For an individual, that can be as simple as not generating twenty versions of something when two will do.

A worked checklist

Here is a checklist you can copy and adapt. It takes about a minute per task.

Before you ask

1. Is there anything personal, confidential or belonging to someone else in what I'm about to paste? If so, remove it or replace it with placeholders.
2. Is this a decision that affects a person (hiring, grading, a health or money choice)? If so, AI can help me think, but it doesn't decide.

While you work

1. Did I ask the tool to show its reasoning or list its assumptions, so I can check them?
2. Does the answer treat people fairly, or does it rely on stereotypes?

Before you use the result

1. Have I checked every name, number, date, quote and link against a reliable source?
2. Would the people relying on this expect to know AI was involved? If yes, I say so.
3. Am I comfortable putting my own name to this?

To see it in action, imagine a small-business owner who wants help writing a reference letter for a former employee. Item 1 catches the first problem: the draft prompt includes the employee's home address and a health issue that explained a long absence. Neither belongs in the prompt, and the health detail doesn't belong in the letter either. The owner rewrites the request with only job title, dates, duties and strengths.

Help me write a one-page reference letter for a former employee. Role: office coordinator, March 2020 to October 2023. Strengths: organized the move to a new booking system, trained three new staff, very reliable with clients. Keep it warm and factual. Don't invent achievements; if you need more detail, ask me.

Item 5 then applies to the result. If the draft says the employee "increased bookings by 30%" and nobody told it that, the number comes out. Item 7 is the final test: the letter goes out under the owner's signature, so every sentence must be one they can stand behind.

Why personal habits matter

Rules for companies take years to write and enforce. Your habits apply today. They also protect the people who never chose to be part of your conversation with an AI: the colleague whose email you might have pasted, the customer whose complaint you summarized, the student whose essay you asked it to grade. Thinking about them is the core of responsible use.

What is UNESCO asking of governments and companies?

The 2021 Recommendation asks member states to build the legal and policy framework that makes ethical AI possible. It calls for stronger protection of personal data, giving people more transparency and control over how their data is used. It explicitly rules out using AI for social scoring and mass surveillance. It covers transparency, accountability and privacy across policy areas such as data governance, education, culture, work, health and the economy, and it asks the people who build and use AI to favor efficient methods that limit environmental harm. UNESCO says it will help countries apply the text and expects regular progress reports from them.

In 2023, as tools like ChatGPT spread into classrooms, UNESCO added more specific requests. In September it published its first guidance on generative AI in education and research. Among other steps, it asks governments to regulate these tools in schools, protect users' data privacy and put limits in place for children under 13. It also asks for training for teachers and protection for the rights of teachers and researchers. UNESCO noted that most countries have no national rules for generative AI yet, and that when it surveyed over 450 schools and universities, fewer than one in ten had formal guidance.

For companies, the message running through both documents is that decisions affecting many people should be fair, open to scrutiny and possible to challenge. For individuals, the personal checklist above is the same message at the size of one desk.