The questions to ask before using AI in schools

A student types a question into an AI chatbot: "Write my essay about the history of Dahomey." Ten seconds later, a neat, confident essay appears. It sounds impressive. It has an introduction, three sections and a conclusion. It also contains two dates that are wrong and a king who never existed.

The student submits it. The teacher is suspicious but can't prove anything. Meanwhile, another student uses the same tool to get a patient explanation of a maths problem she's been stuck on for a week — and finally understands it.

That's AI in schools today: powerful, helpful, risky and confusing — sometimes all at once.

Before a school adopts AI tools, what questions should teachers, school leaders and parents ask?

For background on AI in Africa, read How AI Is Changing Africa: Promise, Risks and the People Building It.

Why this matters now

AI tools — especially generative AI chatbots that write text, answer questions and create images — spread extremely fast after 2022. Students often discovered them before teachers or schools had rules.

UNESCO surveyed schools and universities globally in 2023 and found that fewer than 10% had developed formal policies or guidance on generative AI. UNESCO later published guidance recommending, among other things, a minimum age of 13 for the use of generative AI tools in classrooms, and a strong focus on human-centred use.

Fewer than 10% of schools and universities surveyed had formal guidance on generative AI. (UNESCO global survey, 2023) Most schools are making decisions without a plan. These questions can help.

Question 1: What learning problem are we trying to solve?

Start with the problem, not the tool. "Everyone is using AI" isn't a reason. Ask:

• Do students need more practice and feedback in maths?
• Do teachers need help preparing lessons or translating materials?
• Do students need support in reading or languages?

If AI doesn't clearly help with a real learning problem, it may be a distraction.

Question 2: Will this help students think — or think for them?

Learning happens when students struggle productively: trying, failing, explaining, revising. If AI does all the thinking, students may get answers without learning.

Good uses:

• Explaining a concept in different ways.
• Giving practice questions and hints.
• Giving feedback on a draft the student wrote.
• Helping brainstorm ideas that students then develop.

Risky uses:

• Writing entire essays or homework.
• Solving problems without explanation.
• Replacing reading with summaries.

Joke break: A student asked an AI to write a poem about honesty for his homework. The teacher gave him full marks for the poem — and zero for honesty.

Question 3: How will we handle cheating?

Cheating isn't new — students have copied from friends and the internet for years. But AI makes it faster and harder to detect.

Important: AI detection tools are unreliable. They can falsely accuse students, especially those writing in a second language. Schools should be careful about punishing students based on detector results alone.

Better approaches:

• Design assignments that require personal experience, local context or in-class work.
• Ask students to show their process: drafts, notes, oral explanations.
• Use more in-class writing and oral presentations.
• Create clear rules: when AI is allowed, when it isn't, and how to declare its use.

See Learning beyond exams: projects children remember for assignment ideas that are harder to fake and more meaningful.

Question 4: Is the information accurate?

AI chatbots can produce confident, well-written answers that are wrong. This is sometimes called "hallucination." It's especially common with:

• Local history, names and places.
• Recent events.
• Statistics and citations (AI sometimes invents sources).
• Topics in African languages or African contexts that are less represented in training data.

Students must learn to check AI answers against reliable sources. This is a critical skill — see A practical way to check a claim about Africa and How to read a health claim shared online.

"The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." — Edsger W. Dijkstra, computer scientist

Question 5: Whose data, and where does it go?

Many AI tools collect what users type. Ask:

• What student data does the tool collect?
• Where is it stored?
• Is it used to train the AI?
• Can students use it without creating personal accounts?
• Does it comply with national data protection laws? (Benin, for example, has provisions on personal data protection in its Digital Code, and an authority for personal data protection.)

Children's data deserves special protection. Never require students to enter sensitive personal information into AI tools.

Question 6: Does it work in our languages and context?

Most AI tools work best in English and a few other major languages. They often perform poorly in African languages like Fon, Yoruba, Wolof or Hausa — and may misunderstand African contexts.

African researchers and organisations are working to change this. Communities like Masakhane and companies like Lelapa AI focus on African-language AI. But schools should test tools in their real context before adopting them.

Ask: Does it understand our curriculum? Our local examples? Our students' language? See The role of local languages in health communication for why this matters.

Question 7: Is it fair to all students?

If only some students have phones, internet or paid AI subscriptions, AI can widen inequality. Ask:

• Can all students access it equally?
• Does it work on basic devices and slow internet? (See How to design useful technology for low connectivity.)
• Does it show bias against certain groups?

AI systems learn from data, and data can contain bias. Teach students to notice when AI answers reflect stereotypes.

Question 8: What role will teachers play?

AI should support teachers, not replace them. Teachers bring things AI can't: relationships, care, judgement, knowledge of each student, cultural understanding and motivation.

AI can help teachers:

• Prepare lesson ideas and practice questions.
• Translate or simplify materials.
• Give first-draft feedback that teachers review.
• Save time on admin tasks.

But teachers need training to use AI well — and to understand its limits. Ask: What training will teachers receive? Who will support them?

Question 9: How much does it really cost?

AI tools may seem free, but costs include:

• Devices and internet data.
• Electricity.
• Paid subscriptions.
• Teacher training time.
• Technical support.

Ask whether the same money could achieve more through other investments: books, teacher training, science kits, libraries. See The people keeping public libraries alive.

Question 10: How will we know if it's working?

Before starting, decide how to measure results:

• Are students learning more? (Test scores, quality of work.)
• Are students more motivated?
• Are teachers saving time?
• Are there negative effects (cheating, dependence, privacy issues)?

Start with a small pilot in a few classes. Measure. Adjust. Then decide whether to expand. See How to measure whether an initiative helped.

Teaching AI literacy

Whether or not schools use AI tools directly, students need AI literacy:

• What AI is and isn't.
• How it's trained and why it can be wrong.
• How to use it responsibly and honestly.
• How to check its answers.
• How it affects jobs, privacy and society.

AI literacy is becoming as important as reading and maths. Coding and computational thinking help too — see What a first coding lesson can teach a child.

A simple school AI policy (starter)

A basic policy might say:

• AI may be used for explanations, practice and brainstorming when the teacher allows.
• Students must declare when and how they used AI.
• Submitting AI-generated work as your own is not allowed.
• Students must not enter personal or sensitive information into AI tools.
• AI answers must be checked against reliable sources.
• Teachers will receive training and guidance.
• The policy will be reviewed each year.

A classroom activity: "Catch the AI"

Here's a fun way to teach AI literacy (for ages 13+). The teacher asks an AI tool a question about local history or geography — for example, "Who founded the city of Porto-Novo?" or "What are the main rivers of Benin?" Students then check the answer using textbooks, library books, trusted websites or interviews with elders.

They score the AI:

• What was correct?
• What was wrong or invented?
• What was missing?
• How confident did it sound?

Students quickly learn that sounding confident isn't the same as being right. It's one of the most important lessons of the digital age — and it's also great for learning local history.

Better questions, better choices

AI will shape the world today's students grow up in. Schools can't ignore it — but they shouldn't rush into it either. The best approach is careful, curious and human-centred: ask good questions, start small, protect students and keep teachers at the heart of learning.

After all, the most important thing a school can teach isn't how to get answers from a machine. It's how to ask good questions.

Keep reading: Why African children should learn to build robots, technology and code, The skills students learn when they teach each other, and our Technology & Innovation section.

Questions

Should schools allow students to use AI?

It depends on the learning goal. Many experts suggest allowing AI for explanations and practice with clear rules, while preventing it from replacing students' own thinking and work.

What age should students use generative AI?

UNESCO's guidance suggests a minimum age of 13 for using generative AI tools in classrooms.

Can AI detection tools catch cheating?

They are unreliable and can falsely accuse students, so schools shouldn't rely on them alone.