Education has been suffering from an old wound for decades: the digital gap. First, it was whether or not you had a computer at home; then it was whether or not you had a stable connection. Today, the problem has become more complex. It is no longer just about who connects, but about who can use, pay for and benefit from the new artificial intelligence tools. The dividing line shifts: from device, electricity and WiFi to algorithms and premium subscriptions.
The World Bank has just documented this in a report titled 100 Student Voices on AI and Education. They spoke with students from ten countries (from Cameroon to Rwanda, including Mexico, Peru, and Colombia), to hear how they are experiencing the arrival of AI in classrooms and, above all, in their study routines. The conclusion: artificial intelligence can be an opportunity to democratise knowledge, but also a threat that entrenches inequalities.
Thus, while some students report that ChatGPT or similar tools have become permanent tutors, available 24 hours a day, others can only read about them in headlines, because the reality is that in their countries the internet is either too expensive or too unstable.
The inevitable question that hangs over the entire World Bank’s report is the following: Will AI be the great equaliser of global learning or the new dividing line in education?
AI as an “equaliser” and as a “divider”
In general, most students welcomed the advent of artificial intelligence with enthusiasm. “A tutor for each student, accessible 24 hours a day,” proclaim its advocates. And indeed, for many students, this is true. In the World Bank report, a young man from Indonesia explains that he practises English by chatting with a chatbot, something previously reserved for those who could afford private lessons. In Nigeria, another confesses that it was thanks to AI that he finally understood fluid physics: “It’s like having a private tutor, but free of charge and without pop quizzes”.
So far this is the bright picture. But, as with almost all technologies, equality lasts until the “premium” button appears. Free access exists, yes, but with limitations. Vaguer answers, longer waiting times, system crashes… To get more complete and reliable explanations, you have to pay. And not everyone can afford paying. A Cameroonian student sums it up bitterly: “Those who have a credit card and a good connection speed ahead in a Ferrari; we are still pedalling on bicycles.”
The curious thing is that the students themselves perceive this duality. They speak of AI as a poisoned gift: it opens up opportunities previously unimaginable, but unevenly distributed. In Rwanda, a group of university students recognise that AI allows them to translate inaccessible academic texts, but their mobile data bills skyrocket to the point where they have to choose between studying and eating. In contrast, their European or North American peers hardly think about it at all: for them, the issue is not access, but learning to use the tool critically.
The result is a mix of enthusiasm and frustration. AI appears as the great equaliser in global discourses, but on a day-to-day basis it threatens to become a sophisticated divider, offering brilliant answers to the privileged student and leaving others with the cut-down version of the technological revolution.
Barriers to access in low-income contexts
Artificial intelligence can be a luxury even in its free version. In Cameroon, a student interviewed by the World Bank reported that, in order to use ChatGPT, he had to calculate how many megabytes he had left on his data plan before daring to make a query. “Sometimes I’m afraid to waste the internet on a bad
answer,” he said. It’s not that AI is inaccessible, it’s that it’s expensive. In a country where a gigabyte can cost up to 10% of daily income, every question to the chatbot becomes a small leap of faith.
In Rwanda, the constraints are more prosaic (but no less restrictive for being prosaic): computer labs, where they exist, are full of computers that were already old before the pandemic. “The fans make more noise than the teacher,” quipped one student. Here, AI is a pipe dream that you read about in international reports but does not exist in the real classroom, where the urgency is simply getting the machines to start.
Ethiopia adds another layer: teachers. According to several testimonies, older teachers hardly approach AI tools, either out of fear of losing authority or through sheer lack of knowledge. The problem is not only the lack of internet or equipment, but also the lack of training support. An Ethiopian student summed it
up like this: “They tell us to use AI, but they don’t know how to guide us. It’s like learning to drive with someone who has never driven a car before.”
Meanwhile, in more privileged contexts, the conversation revolves around other dilemmas: how to teach AI critically, how to avoid plagiarism, how to assess when the student has an assistant who responds instantly. The contrast is stark: some struggle to load a page without exhausting their data plan; others argue whether asking the chatbot for an entire essay is cheating or simple efficiency.
Barriers to access, far from being anecdotal, map a new inequality. Artificial intelligence not only amplifies the gap between rich and poor, but introduces a perverse nuance: those who could benefit most from a digital tutor are precisely those who can least afford to pay for it or even connect to it.
AI appears as the great equaliser in global discourses, but on a day-to-day basis it threatens to become a sophisticated divider, offering brilliant answers to the privileged student and leaving others with the cut-down version of the technological revolution.
Implications for educational equity
Artificial intelligence promises to personalise learning and open the door to quality contents on a global scale. But, as students point out, that promise is not within everyone’s reach. Those who already had an advantage (better devices, better connection, more cultural capital) now have an “intelligent assistant” that multiplies their opportunities. The others are still waiting for the lab computer to switch on. This means that AI, rather than levelling out, could be widening the existing gap.
The World Bank collects testimonies from young people who fear being left behind in the labour market. “If AI is the future of work, what about those who can’t use it?” asked a student from Rwanda. The concern is not minor: the OECD estimates that in the next ten years, up to 30% of jobs will require skills
linked to AI or automation. If preparation for these tasks is limited to a connected elite, inequality will be entrenched not only in the classroom, but also in the economy.
UNESCO, in its report AI and Education: Guidance for Policy Makers (2019), warned that AI could become an “inequality multiplier” if not accompanied by inclusive policies. What in high-income countries is discussed as an ethical problem (i.e., How to avoid plagiarism? How to teach critical thinking?) in resource-poor contexts is a much more basic problem: access.
The ethical dilemma is obvious. Is it fair to assess students who have a digital assistant alongside those who can barely open a browser? Can a scenario where half of the classroom has a virtual tutor and the other half does not even have WiFi be called equal opportunities? AI has not invented inequality, but it is
reshaping it with a technological veneer that makes it less visible and therefore more dangerous.
Ways to close the gap
The AI gap is not written in stone. It can be mitigated. How? The first step is obvious: infrastructure. Without affordable internet and suitable devices, any discussion of AI ethics borders on the cynical. The World Bank insists that investment in Internet connection (conectivity) is as urgent as teacher training.
The second step has to do with training. Schools and universities must teach not only how to “use” AI, but also how to understand it critically: how algorithms work, what biases they carry, what limits they have. Otherwise, the tool becomes a black box that few know how to interpret. UNESCO calls it “AI
literacy” and puts it on the same level as reading and mathematics.
The third front involves technology companies. It is not enough to launch free versions as a commercial hook. Inclusive models are needed: plans that work with bad Internet connection, affordable prices for vulnerable contexts, tools designed with linguistic and cultural diversity in mind. The OECD warns that if
innovation is concentrated in wealthy markets, inequality will be inevitable.
In other words: closing the gap is a political, pedagogical and business effort. And the sooner it starts, the better. Because every academic year that passes without solutions multiplies the risk that AI will cease to be the great equaliser and become the new wall of education.
