Teach Both, Test Both: What Our Children Need to Flourish in the AI Era

Children should not be prepared for an AI-free world that will not exist. They should not be made dependent on AI for abilities they still need to own. The educational principle I find most useful is simple: teach both and test both.

Economist Daniel Susskind, writing on 6 September 2026 as a father of three, argues that children need strong foundations alongside intelligent use of new tools. His framing connects directly with my experience as a mother of three and as a student of psychology: development is not only about the final answer. It is about building the mind that can understand, question and use the answer.

What “teach both” means

A child should learn to calculate and also use a calculator. Write and also use an editor. Read a complete text and also ask an AI tutor for explanation. Search independently and also learn how an assistant synthesises sources.

These are not contradictory methods. One builds internal capability; the other extends capability. Problems emerge when extension arrives before foundation or when foundation is taught as if tools do not exist.

What “test both” means

Schools need assessments with different purposes. An unaided assessment asks what knowledge and skill the child can independently retrieve and apply. An AI-inclusive assessment asks whether the child can frame a problem, use tools responsibly, verify outputs and produce something valuable.

• Without AI: literacy, numeracy, memory, reasoning, handwriting, explanation and foundational knowledge.
• With AI: prompting, source checking, comparison, revision, collaboration and responsible tool selection.
• Oral explanation: can the child describe the work in their own words?
• Practical demonstration: can the child apply knowledge outside the exact format in which it was taught?
• Reflection: can the child identify what the tool helped with and what still required human judgement?

Attention becomes a core skill

An AI system can generate endless stimulation, alternatives and instant responses. Children also need the ability to remain with one difficult problem, tolerate uncertainty and complete a task before seeking novelty.

Attention should therefore be taught like reading: through graduated practice, supportive environments and realistic expectations. A brain break can restore cognitive energy. A mind break can create emotional space. Neither should become avoidance of every demanding moment.

Personalisation without isolation

AI tutors can explain a concept repeatedly, change the level of difficulty or provide another example without embarrassment. This could be valuable for children who need different pacing or communication routes.

But education is also social. Children learn turn-taking, negotiation, empathy, repair, shared attention and belonging from humans. Personalised software must not become an excuse to withdraw teaching assistants, specialist support or peer relationships.

Do not future-proof a child into one job title

Technology changes job tasks faster than families can predict a single permanent career. Education should develop the ability to recognise important problems and learn the tools required to address them.

Foundational literacy and numeracy remain powerful because they support almost every future route. So do curiosity, creativity, ethical reasoning, communication and the confidence to learn again.

A family AI agreement

1. Name the purpose. Explain whether AI is being used to understand, practise, create or verify.
2. Try first when appropriate. Give the child a chance to form an initial answer before assistance.
3. Never hide uncertainty. Show that AI can be wrong and that adults verify too.
4. Protect private information. Children should not enter personal, medical or school-identifying details without informed adult control.
5. Preserve offline competence. Continue books, play, conversation, practical tasks, movement and handwriting.
6. Create together. Use AI as a family learning activity rather than always as a solitary screen.
7. Review the effect. Ask whether the tool increased understanding or merely finished the task.

The role of parents and teachers

Adults do not need to know every new model. We need enough literacy to ask good questions, establish boundaries and remain emotionally available. Children watch whether we use AI to think more carefully or simply to avoid effort.

Today’s research on AI and linguistic diversity adds another reason for involvement. A child’s developing voice should not be polished into generic adulthood before it has had time to become itself.

The labour-market discussion matters too. China is already measuring AI’s effect on jobs. Education must prepare children for transition without making anxiety the centre of childhood.

Teach children what they can do alone. Teach them what they can do with AI. Then teach them how to know which kind of strength the moment requires.

Sources and date note

This article reflects information available on 6 September 2026. Read Daniel Susskind’s 6 September essay for parents. This MaryChuks Sunday Premium develops an independent psychological and family framework.