When chatbots deliver instant answers, they eliminate the “productive struggle” kids need to learn how to think. But in overcrowded African classrooms with severe teacher shortages, the alternative to an AI tutor is often no tutor at all. So, where should Africa draw the line? In his episode of Delve Into AI, ADONIJAH NDEGE dives into why copying New York’s ban would be a mistake, the dangerous trap of cognitive shortcuts, and why the future of AI in African education depends on one radical shift: letting the machine ask the questions while the student generates the answers.
On September 3, New York City, home to the largest public school system in the United States, banned the use of artificial intelligence (AI) in classrooms through the eighth grade. The decision brought to mind a conversation I had with Kurt Gammelgaard Nielsen, chief executive of EdTech Denmark, over dinner in Copenhagen on August 25.
I had mentioned that I write Delve Into AI, a weekly column for TechCabal that explores how artificial intelligence is reshaping life and work across Africa. Naturally, our conversation turned to what AI might mean for education, particularly for children in their formative years.
Nielsen’s concern was not whether students should eventually learn to use AI—they almost certainly will have to—but whether introducing tools designed to simplify thinking too early could interfere with the difficult, and sometimes frustrating, process through which children actually learn to think.
“AI should augment human learning, not replace it,” he told me. “Children should be empowered to think critically, explain deeply, and learn more effectively.”
A school is not a factory
This distinction may be the most critical element in the debate over AI in education. Many conversations focus on what AI can do for students, teachers, and schools. For Africa, where access to quality education in remote areas remains a persistent challenge, the primary question often centres on how technology can bridge the gap.
However, my conversation with Nielsen suggested that we might be asking the wrong question entirely. Before asking what AI can do for students, we must first ask what education is meant to accomplish for them.
“Does it support learning or does it create cognitive ‘offloading’?” he posed. “AI won’t make us smarter by default. Finished answers given by chatbots replaces reasoning.”
Across the Nordic countries, early enthusiasm for classroom technology is colliding with growing concerns over its long-term impact on cognitive development, attention spans, and child safety. It is a tension that has now crossed the Atlantic, leading New York Mayor Zohran Mamdani to conclude that, during a child’s foundational years, technology may carry too high a developmental cost.
Nielsen speaks in a panel discussion on Translating research into commercial ventures through university spinouts on August 27, 2026, during the Nordic-Africa Summit at TechBBQ. Image source: IntroAfrica/ Mikkel Becker-Aakervik
A school is not a factory whose primary objective is to produce correct answers as efficiently as possible. Childhood learning is deliberately inefficient—and should remain that way.
A child reads a sentence repeatedly as comprehension builds. They write and rewrite individual letters because physical writing helps organise thought. They struggle with basic arithmetic before receiving a calculator, and they memorise facts even though search engines can supply them in milliseconds.
Like Nielsen and Mamdani, I believe this initial struggle during formative years forms the very bedrock of learning.
Generative AI tools like ChatGPT, Grok, and Deepseek present an almost opposite proposition. Their core promise runs counter to the process of learning: ask a question, and the machine produces a finished essay. Photograph a maths problem, and it instantly generates the solution.
“As a student, knowing what problem you are trying to solve remains more important than having an AI capable of solving it,” he said. “Without the capacity to explore, analyse, and accurately define problems, efforts may be misdirected at incorrect issues.”
Built for those who already have skills
For an adult equipped with foundational knowledge, AI is a powerful multiplier. I use AI regularly, as do many readers of this column. Delegating routine tasks frees up time for critical judgment, strategy, and creative execution.
However, a 10-year-old in primary school is still developing the baseline cognitive architecture upon which those shortcuts depend.
This is why researchers like Nielsen distinguish between AI for productivity and AI for learning. The former asks whether a task was completed faster; the latter asks what fundamental shift occurred in the learner after the task was completed. The Organisation for Economic Co-operation and Development (OECD) reached a similar conclusion. Its 2026 Digital Education Outlook warned that general-purpose generative AI can improve the superficial quality of the work students produce without necessarily improving what they actually comprehend.
But the problem may not be entirely AI; it may be what we ask the technology to optimise for. Nielsen put it to me another way: “The AI asks, the student generates.”
If you think about it, it is a radical idea that could lead to different outcomes in AI for learning. Most chatbots today work in precisely the opposite direction. Humans ask questions, and then the machine gives answers.
In a properly designed educational system, the machine might instead ask a child why they arrived at that answer. It could challenge an assumption, offer a clue rather than a solution, ask the learner to explain something in their own words, or require the learner to develop an argument before writing an essay.
That way, the learner remains intellectually responsible for the work. And there is evidence that this kind of design can work. A 2025 Harvard experiment involving undergraduate physics students found that an AI tutor produced substantially greater learning gains than an active-learning classroom.
But this was an engineered system grounded in pedagogical principles that guided students through the material rather than simply completing tasks for them. It is different from the approach taken by existing tools.
Africa has more to gain, and more to lose
It is easy to look at New York and conclude that African governments should follow its lead. I am not convinced that would be wise. The conditions under which children learn vary across the continent, but in many African countries, the constraints facing schools are very different from those in New York.
A child in Manhattan may have access to well-resourced schools, libraries, and books, as well as tutoring outside the classroom. In large parts of Africa, the alternative to an AI tutor may be no tutor at all.
That changes the calculation dramatically. In eastern and southern Africa, only about 11% of children in countries targeted by a 2025 World Bank learning programme can read and understand a simple text by age 10. Across sub-Saharan Africa, learning poverty is about 89%.
Most classrooms in Kenyan public schools, for example, are overcrowded, with some having one teacher facing 60 children without enough learning materials. There are rural communities, such as those in Northern Kenya, where teachers are scarce. AI could help solve some of those problems.
A teacher helping learners at Tom Mboya Primary School, Dandora Nairobi. Most classrooms in Kenyan public schools are overcrowded. Image source: UNICEF
A properly designed system could provide a child with unlimited practice in mathematics, help a teacher generate exercises at varying levels of difficulty, and translate explanations into local languages. It could also help teachers identify patterns in a pupil’s performance that suggest the pupil is falling behind, though that would depend on the system’s quality, the data it has access to, and the infrastructure supporting its use.
The United Nations Educational, Scientific, and Cultural Organisation (UNESCO) argues that generative AI could strengthen foundational learning in sub-Saharan Africa when deployed alongside proven teaching approaches rather than as a substitute for teachers. The opportunity includes low-cost devices, local-language learning, faster assessments, and tools that help teachers adapt material to a pupil’s actual level.
This leaves African governments with a more complicated choice than simply deciding whether AI belongs in classrooms. The question is what kinds of AI should be allowed, at what ages, for which tasks, and under whose supervision. Keeping the technology entirely out of schools could mean forgoing tools that might expand access to learning support. But introducing general-purpose chatbots without clear pedagogical goals could simply make it easier for children to obtain answers without developing the skills required to reach them.
That risk may be particularly consequential in education systems already struggling with foundational learning. Where guardrails around AI remain weak, handing children powerful chatbots without first deciding what they should—and should not—do could compound existing learning problems rather than solve them.
Perhaps Nielsen’s simplest argument is worth remembering: let the AI ask, and the student generate. The great promise of AI in African classrooms should be that they will struggle with better questions, teachers and tools around them.
The machine can eventually help them move faster. But in the formative years, we should first give children the time, space, and frustration they need to learn to think for themselves.
True scale demands moving beyond surface-level integrations to robust execution. We’ve filtered the noise out of Moonshot 2026, optimising the conference strictly for high-calibre connections between startup founders, global financial operators, enterprise leaders and individuals rewiring Africa’s technical frameworks. Get 20% off Early Bird tickets for a limited time.
from TechCabal https://ift.tt/nNd6OXj
via IFTTT
I am the CEO of 9JAGISTLAND.My name is Adedeji Timothy and my co-owner is Oyekola Damilare.This blog is born out of our passion to educating to people in all aspect of internet life.We are enthusiastic about writing posts and articles online.
Write your views on this post and share it. ConversionConversion EmoticonEmoticon