Can AI Tackle Learning Poverty In The Global South

A stark number sets the stakes: seven in ten 10-year-olds in low and middle income countries cannot read a simple sentence. We take that reality out of the abstract and into a crowded classroom, following Saad, who is lost in long division, and Fatima, who is bored because the pace is too slow. F

rom there we explore whether AI can truly help systems leapfrog toward quality education, or whether it risks becoming a shiny diversion that deepens inequality.

TLDR / At A Glance:

• learning poverty at 70 percent among 10-year-olds in low and middle income countries
• web of exclusions across gender, disability, conflict, language and culture
• access success but quality failure in crowded classrooms
• personalised AI tutoring that diagnoses gaps and adapts tasks
• high-dosage tutoring gains in Edo State, Nigeria
• teacher workload relief through planning and grading automation
• Nova Sola WhatsApp chatbot saving one hour per lesson plan
• local language content generation to counter colonial curricula
• universal AI literacy for critical, ethical use
• co-intelligence as a design goal and last-mile inclusion

We dig into concrete, on-the-ground examples. An after-school pilot in Edo State, Nigeria used an AI tutor to deliver learning gains equal to one-and-a-half to two years in only six weeks, showing what high-dosage, one-on-one support can do when cost barriers fall. We look at teacher-centred tools too: a WhatsApp-based lesson planning assistant in Brazil that saves an hour per plan, turning automation into time for rest, feedback, or one-on-one care. And because connectivity is the fault line, we unpack “AI unplugged”: paper tests photographed on a single phone, uploaded later, analysed in the cloud, and returned as simple, actionable diagnostics that guide tomorrow’s lesson. We also spotlight the urgent need for culturally relevant content, highlighting rapid generation of children’s books in local languages to replace decades-long shortages.

But speed without equity is a trap. We name the Matthew effect at play when solutions assume electricity and broadband that most schools do not have. 

We weigh innovation against transformation, asking not only how to teach but what to prioritise when labour markets shift and community knowledge matters. 

Alongside sobering OECD futures like “education outsourced,” we argue for universal AI literacy so every child can question sources, spot bias, and understand how recommendations are made. The north star is co-intelligence: humans leading, AI extending reach, with system design that includes infrastructure, teacher training, governance, and language.

If you care about closing the learning gap without creating a permanent underclass, this conversation is for you. 

Listen, share with a colleague who works in education or development, and leave a review telling us one low-tech idea that could scale in your context.

 Your feedback helps more people find the show and keeps this work moving forward.

Transcript

The Learning Poverty Shock

Google Agent 1 0:00

Aaron Powell It's a gut punch of a number, isn't it?

Google Agent 2 0:02

It really is. I mean, we're not talking about 70% of kids failing, you know, a calculus test.

Google Agent 1 0:07

No, not at all.

Google Agent 2 0:08

We're talking about 70% of 10-year-olds in low and middle-income countries unable to read and understand a simple text.

Google Agent 1 0:15

Aaron Powell A simple sentence, the sun is hot, things like that. They just can't process it. Trevor Burrus, Jr.

Framing AI Beyond The West

Google Agent 2 0:20

So the World Bank has a name for this, learning poverty. And honestly, it just feels like this silent catastrophe. We spend so much time worrying about the future of work, but if that foundation is missing, the whole thing just collapses.

Google Agent 1 0:36

And that uh that stark reality is really the starting point for today's deep dive. We're getting into a source called Artificial Intelligence and Education in the Global South, a systems perspective.

Google Agent 2 0:47

Aaron Powell And I'm so glad we are, because when we usually talk about AI and education, the conversation, it feels very Western.

Google Agent 1 0:55

It does completely. The headlines are all about, you know, did Chat GPT write this college essay or is an algorithm going to crash the stock market? It's very high income anxiety.

Google Agent 2 1:03

Aaron Powell Exactly. But this source forces us to pivot.

Google Agent 1 1:07

Yeah.

Google Agent 2 1:07

It asks a much bigger, much higher stakes question. Can AI solve that 70% problem? Can it fix this global learning crisis where you have millions of kids physically sitting in a school but learning, well, nothing?

Google Agent 1 1:20

Aaron Powell Or, and this is the other side of the coin, is it just a shiny new distraction?

Google Agent 2 1:25

Right.

Google Agent 1 1:26

Is it just going to be a toy for the wealthy that actually widens the gap and leaves the poor even further behind? That's the real tension here.

Web Of Exclusions

Google Agent 2 1:33

Aaron Powell And to unpack that, we have to look at where 90% of the world's under 18 population actually lives.

Google Agent 1 1:39

Aaron Powell The Global South.

Google Agent 2 1:41

So let's start there. When the authors say global south, they're not just pointing at a map, are they?

Google Agent 1 1:45

Aaron Ross Powell No, not really. It's more of an economic shorthand. We're talking about regions in Africa, Latin America, developing Asia, the Middle East. And the key thing for this discussion, anyway, is resource scarcity. I mean, the resources per student are just dramatically lower than in the global north.

Google Agent 2 2:00

Aaron Powell But the source uses a phrase that I found really powerful. They don't just talk about a lack of money, they talk about a web of exclusions.

Google Agent 1 2:07

It's a great image, isn't it? Because it suggests it's not one thing, it's a system. Think of it like a net. In a well-resourced system, the net catches everyone. But here, the net has huge holes and it's actively filtering kids out.

Google Agent 2 2:21

So what are the strands in that web?

Google Agent 1 2:23

Well, gender is a massive one. In sub-Saharan Africa, for every 100 boys of primary school age who are out of school, there are 123 girls.

unknown 2:35

Wow.

Google Agent 1 2:36

The system is just rigged against them from the start.

Google Agent 2 2:38

And then you add in things like disability or conflict. The source says half of the world's chronically out-of-school kids live in conflict zones. It's not, I don't have a textbook. It's my school might not be there tomorrow.

Google Agent 1 2:49

And yet, and this is a really important paradox for you to grasp to see why AI is even part of this conversation. Enrollment has actually skyrocketed.

Google Agent 2 2:57

Right, something like 90% globally for primary school. The kids are in the building.

Access Won, Quality Lost

Google Agent 1 3:02

The kids are in the building. We spent decades solving the access crisis. We built the schools, we got them in the door, but we completely failed to solve the quality crisis. They're sitting there, but as we said, they're not learning.

Google Agent 2 3:14

Which brings us to the big idea the leaffrog.

Google Agent 1 3:17

The leapfrog.

Google Agent 2 3:18

The idea that maybe, just maybe, these countries don't have to take the same slow industrial age path the West did. They can use AI to jump straight to something better.

Saad And Fatima’s Classroom

Google Agent 1 3:29

That's the hope, the techno-optimist view, that technology can compress decades of development into just a few years. But to see if that's real, we have to zoom in. We need to get out of the 30,000-foot view and into an actual classroom.

Google Agent 2 3:43

So let's go to one. The source paints this picture of a classroom in Multan, Pakistan. It's hot, it's crowded, maybe 50, 60 kids. And we meet two students, Saad and Fatima.

Google Agent 1 3:54

And this scenario is the perfect encapsulation of what they call the one size fits all failure.

Google Agent 2 3:58

Okay, so walk us through it. What's happening with Saad?

Google Agent 1 4:00

Saad is just lost. The teacher's at the front explaining, let's say, long division, but she's got 50 other kids, so she's teaching at one pace, the pace of the curriculum. For Saad, that's way too fast. He missed a key step three weeks ago, and now nothing makes sense.

Google Agent 2 4:17

Oh, I think we've all felt that. That panic where you're just praying the teacher doesn't call on you. You just want to become invisible.

Google Agent 1 4:22

Exactly. It's shame. So he's hiding. He's totally disengaged. Now look a couple rows over at Fatima.

Google Agent 2 4:29

She has the opposite problem.

Google Agent 1 4:30

The exact opposite. She got this concept ten minutes ago. She's bored out of her mind. She's doodling. She's disengaged because the system is holding her back.

Google Agent 2 4:39

And the poor teacher is caught in the middle. What can she even do?

Google Agent 1 4:42

She's trapped. She slows down for Sa'ad. She loses Fatima and 10 other kids. She speeds up for Fatima. She loses Saad and 20 others. So she does the only thing she can. She teaches to the middle.

Google Agent 2 4:53

And loses almost everyone.

Google Agent 1 4:55

And loses almost everyone. This is where AI comes in. Not to replace the teacher, but to break that model.

Google Agent 2 5:01

To bring in personalization.

Personalised AI Tutoring

Google Agent 1 5:02

Right. Personalization. It sounds like a buzzword, I know. But for Saad, it looks like this. He gets on a tablet, maybe in a shared computer lab. The AI tutor sees, hey, Saad is stuck on division. But it doesn't just repeat the same lecture.

Google Agent 2 5:17

It tries a different angle.

Google Agent 1 5:18

Exactly. Maybe it shows him a visual diagram, or it breaks it down into a tiny micro step. It goes all the way back to subtraction to make sure he has that foundation. It meets him right where he is.

Google Agent 2 5:28

And for Fatima.

Google Agent 1 5:29

It pushes her. It says, Great, you got this. Here's a harder problem. Here's a logic puzzle. It keeps her challenged and engaged.

Google Agent 2 5:36

Aaron Powell There was this one example that just blew me away from Edo State, Nigeria.

Google Agent 1 5:41

Yes. This case study is so important because it's not a theory. It actually happened.

Google Agent 2 5:46

So it was an after-school pilot. They used Microsoft Copilot as an AI tutor. And the results, I mean, I had to read the numbers twice.

Google Agent 1 5:54

They are unbelievable. In just six weeks, the students in that program saw learning gains that were equivalent to one and a half to two years of normal school.

Google Agent 2 6:03

Okay, stop for a second. Two years of learning in six weeks. How is that even possible? Is it just that the baseline, the normal school, was so bad?

Google Agent 1 6:13

That's definitely part of it. When the baseline is that low, any effective intervention is going to look miraculous. But it also just shows the power of high dosage one-on-one tutoring. We've always known it's the gold standard. It's just been too expensive. AI makes it cheap and scalable.

Google Agent 2 6:30

And the source said it was especially effective for girls.

Edo State’s Six-Week Surge

Google Agent 1 6:32

Yes, and that gets back to that web of exclusions. A girl might feel less comfortable raising her hand in a crowded class. She doesn't want to look slow, but the AI doesn't judge. It's a safe space to fail and try again.

Google Agent 2 6:44

Incredible. And it wasn't just Nigeria. They mentioned MindSpark in India, Khan Academy in Guatemala. It seems like this really works.

Google Agent 1 6:51

It does, but and it's a huge but we've only talked about the student. We have to talk about the teacher.

Google Agent 2 6:55

Absolutely, because that teacher in Pakistan is facing her own crisis.

Google Agent 1 6:59

The numbers are, frankly, terrifying. In sub-Saharan Africa, the percentage of qualified primary teachers actually dropped. It went from 85% in 2000 down to 65% in 2020.

Google Agent 2 7:12

Aaron Powell That's going in the wrong direction. We're getting more kids into school with fewer qualified people to teach them.

Google Agent 1 7:18

The system is diluting itself. So the conversation can't be AI versus teachers. It has to be how how can AI be an assistant, a force multiplier?

Google Agent 2 7:27

Starting with all the paperwork, the lesson planning, the grading.

Google Agent 1 7:30

That's the low-hanging fruit. The source has this great example from Brazil, a program called Nova Sola.

Google Agent 2 7:36

And what I loved about this is that it used technology everyone already has.

Google Agent 1 7:41

That's the genius of it. You don't need a new app. They built a chat bot right inside WhatsApp. So a teacher, instead of struggling on a Sunday night to write a lesson plan, can just chat with the bot.

Google Agent 2 7:50

I need a lesson on the Brazilian gold rush for fifth graders.

Google Agent 1 7:53

And bang, it gives you a structure, activities, quiz questions. The data showed it saved teachers about an hour per plan.

Google Agent 2 8:00

An hour? Think what a teacher could do with an extra hour.

Google Agent 1 8:03

That's an hour of rest. An hour to talk to a struggling student. That is human-centered efficiency.

Teachers As Force Multipliers

Google Agent 2 8:10

Okay, but I have to push back here because I know what you're thinking. This all sounds great if you have 4G in a smartphone. What about the rural village with no internet?

Google Agent 1 8:19

The infrastructure gap? It's the biggest hurdle. You can't use Chat GPT if you can't get online.

Google Agent 2 8:24

Right.

Google Agent 1 8:24

Or can you? This is where the source talks about a concept I love: AI unplugged.

Google Agent 2 8:29

AI unplugged. Tell me more.

Google Agent 1 8:30

Aaron Powell It's the work of a professor named Sidji Isotani, and his whole point is you have to design for the reality on the ground.

Google Agent 2 8:38

Aaron Powell So what does that look like?

Google Agent 1 8:39

It's high-tech, backend, low-tech user experience. So imagine a classroom, no computers, kids take a test on paper.

Google Agent 2 8:47

Okay. Standard.

Google Agent 1 8:48

The teacher has one smartphone. She just takes a photo of the completed test. Maybe she doesn't have a signal right there. But later, when she goes into town, the photo uploads. The AI and the cloud grades the tests, and this is the magic. It analyzes the patterns.

Google Agent 2 9:02

You can see what everyone got wrong.

Google Agent 1 9:03

Exactly. It sends a simple text back to her phone. Eighty percent of your class failed question three. They don't get subtraction with borrowing. You need to reteach it.

Google Agent 2 9:14

That is brilliant. It uses the high-tech part where it can and the low-tech part where it must.

AI Unplugged For Low Connectivity

Google Agent 1 9:19

The tech adapts to the school, not the other way around. And you can apply that same logic to curriculum.

Google Agent 2 9:25

Right, because so many kids are learning from books written in languages they don't speak at home about cultures they don't recognize.

Google Agent 1 9:31

The colonial legacy problem. You're trying to learn to read and translate at the same time. It's exhausting. But AI can generate content. There was an example from Molly.

Google Agent 2 9:41

Robot Molly.

Google Agent 1 9:42

Yes. They used AI to create over 180 children's books in the local language, Bomra, in less than a year. That would have taken a decade before.

Google Agent 2 9:51

Okay, so the potential is obviously huge. Saad gets a tutor, the teacher gets an assistant, but we have to talk about the risks. The source brings up the Matthew effect.

Google Agent 1 10:01

From the Bible. To those who have, more will be given. From those who have not, even what they have will be taken away.

Google Agent 2 10:08

The rich get richer.

Google Agent 1 10:09

And in this context, the fear is that AI just helps the kids who are already privileged. The ones with good internet and new devices, they'll use AI to accelerate even faster.

Google Agent 2 10:19

While the kids in the villages without power get left even further behind.

Google Agent 1 10:23

And the stats on that are just they're sobering. In sub-Saharan Africa, only 22% of primary schools even have electricity. 28% have internet.

Google Agent 2 10:33

So if your solution requires the internet, you're immediately excluding most of the schools. You're building inequality right into the system.

Google Agent 1 10:39

Aaron Powell That's the nightmare scenario. You just bolt on AI to a broken system and you end up creating a faster, more efficient, but even more unequal system.

Google Agent 2 10:50

Aaron Powell The source also makes this really sharp distinction between innovation and transformation.

Google Agent 1 10:55

Yeah, this is a key point. Innovation is just doing the same thing a little bit better. A digital worksheet instead of a paper one. It's still a worksheet.

Google Agent 2 11:02

It doesn't fundamentally change the process.

Local Language Content At Scale

Google Agent 1 11:04

Exactly. Transformation is reimagining the whole system, asking if AI allows us to change what we teach, not just how. And there are different paths this could take. The OECD laid out four possible futures.

Google Agent 2 11:17

Aaron Powell One is just schooling extended schools. But there was a scary one. Education outsourced.

Google Agent 1 11:27

That's the one that gives me chills. That's where public systems collapse and private tech companies take over. If you can pay, you get a world-class AI tutor if you can't. Tough luck.

Google Agent 2 11:37

A complete bifurcation of society.

Google Agent 1 11:39

Absolutely. And the other two are more hopeful: schools as learning hubs and learn as you go, but that outsourced one is the danger.

Google Agent 2 11:47

Which is why the book argues we need universal AI literacy.

Google Agent 1 11:50

For everyone, not just coders. You see China mandating AI classes from primary school up. Uruguay is focusing on the ethics. Kids need to know how this algorithm works? Is it biased?

Risks And The Matthew Effect

Google Agent 2 11:59

Or they'll just be passive consumers, ruled by the tech instead of using it.

Google Agent 1 12:02

Exactly.

Google Agent 2 12:03

So let's bring it all together. Can AI solve the learning crisis?

Google Agent 1 12:06

The source lands here. It has the potential to solve the Saad and Fatima problem. It has the potential to unburden that teacher in Brazil. But, and this is the most important part, only if you treat it as one piece of a much larger system.

Google Agent 2 12:20

It's not a magic wand you can just wave over a school.

Google Agent 1 12:22

You can't just airdrop a bunch of tablets and walk away. You have to think about electricity, teacher training, cultural relevance, all of it.

Google Agent 2 12:30

And we have to remember the stakes here. It's not just about test scores.

Google Agent 1 12:34

It's about the 251 million children who are completely out of school right now. It's about basic human dignity. If you can't read, your entire life is compromised. The goal isn't just cooler gadgets, it's giving those kids a real fighting chance.

Google Agent 2 12:49

That feels like the right place to land. But I want to leave you with one final thought from the text. It talks about us moving toward a world of co-intelligence, humans and AI working together.

Innovation Versus Transformation

Google Agent 1 13:00

Right. And it says the biggest risk isn't that AI will replace teachers. The biggest risk is that we create a permanent global underclass. A whole group of people who were simply bypassed by this revolution because they happen to live in a place without reliable electricity.

Google Agent 2 13:15

So the challenge for all of us for the next decade isn't just about building smarter AI. It's about building AI unplugged. How do we design these systems to reach everyone, especially those at the last mile? That's something we're thinking about.

Google Agent 1 13:29

It certainly is.

Google Agent 2 13:30

Thanks for diving in with us.