It’s just Tobi and Feyi in the studio this week, diving into a conversation that started with a simple question: How could AI actually transform Nigeria - and Africa more broadly?
I truly enjoyed this podcast. You touched on something very poignant, the schooling-versus-learning crisis currently bedevilling Nigeria.
It’s no exaggeration to say that, even as you alluded to here, despite increased investment in education over the past two decades, Nigeria’s progress has been more about expanding schooling than improving learning. No doubt, enrolment rates have risen significantly under the Universal Basic Education (UBE) programme, and more schools have been built across the country, even in places where classrooms are not needed. (UBEC has said they plan to revise this in their Act; we await the final amendments.)
Yet learning outcomes remain alarmingly low. The World Bank in 2022 estimates that about 70% of Nigerian children cannot read and understand a simple text by age 10. Erratic and inconsistent national assessments also reveal that a large proportion of pupils in upper primary cannot perform basic arithmetic.
Thinking about the assessment issue Tobi mentioned, I realise it's a political minefield. The results are almost always bad news for politicians. Here in my state, for instance, education is tuition-free, which sounds great. But it means that if an exam shows a huge number of students failing at a particular school, the principal is in for a world of pain. To escape the government's rage, teachers and principals are essentially forced to invent better grades for their students.
When Tobi mentioned the 40,000 sacked teachers, I laughed out loud here in the office. Sadly, most states are not sincere about revealing the true staff strength of teachers on their payroll. As someone working in the primary education sector (SUBEB), it is hard to believe that I do not even know the total number of primary school teachers in my own state (O**).
Equally troubling is how this same lack of transparency is extended to development partners who spend huge amounts on our education sector.
The same goes for enrolment numbers. Officers are never allowed to give accurate figures. Two sets of enrolment sheets are always created, one "political" and one "actual." It's a behemoth issue.
Anytime I want to discuss the issue of Cement in any fora, my to go reading has always been your "Konkere" article. 😂. FF and the billionaire.
Thank you for this detailed comment. It really bothers me that we are not taking the Learning crisis seriously. It's like leaving billions of dollars on the sidewalk. A study by Eric Hanushek once estimated that we can increase global GDP by up to $700 trillion just by closing the basic skills gap in education.
I am glad I tuned in for this. Huge fan. This one struck close to work.
What can AI do for Nigeria?
I would hope to answer that question. I must apologize. I am not sure I write that good.
AI is a revolutionary tech, when I say AI I am referring to machine learning neural networks. You provide training data to a neutral network with the expectation that when your model processes data out in the wild, it can correctly provide a relationship based on training data.
This neural networks have been successful in applications such as computer vision, which I have been privileged to work with.
These neural networks has found successful applications also in Large Language Models. So when folks talk about AI these days, most likely refer to LLMs.
However, LLMs aren't going to be impactful such as the Internet or smartphone. At least not yet, there's a handful of challenges bedeviling its potential. Neural networks require incredibly high powered hardware for training and inference. A barrier to scaling to ubiquitous devices such as smartphones and laptops. There are numerous other challenges that I would restrain from listing as I suspect being verbose would obfuscate my point.
AI is next level tech built upon existing infrastructure. Infrastructure that requires high capacity educated population.
AI is already being used in dumber forms, software apps, search engines, etc.
There's nothing new to see if Nigeria hasn't utilized the existing possibilities that tech offers, and it's not really hard to see.
AI is not cheap. It's highly subsidized right now thanks to the hype. The pressure to find the successful utility for AI that would contribute to productivity is still high in the wind.
There are also the hard realities of AI use. We are witnessing in real time in the software space how speed of churning out code doesn't improve existing software products, or help you figure out stuff that people would buy.
I fear that AI compressing the training of teachers would not impact learning outcomes for students.
There's just so much hype right now that it's almost career suicide to correctly assess what is possible with AI, you jump on the hype train or you keep mum. My spicy opinion is there's a plateau right now and there's nothing new to see unless a new breakthrough in the hardware or new methodology arises. The next 1 year would be totally interesting in the AI space.
I truly enjoyed this podcast. You touched on something very poignant, the schooling-versus-learning crisis currently bedevilling Nigeria.
It’s no exaggeration to say that, even as you alluded to here, despite increased investment in education over the past two decades, Nigeria’s progress has been more about expanding schooling than improving learning. No doubt, enrolment rates have risen significantly under the Universal Basic Education (UBE) programme, and more schools have been built across the country, even in places where classrooms are not needed. (UBEC has said they plan to revise this in their Act; we await the final amendments.)
Yet learning outcomes remain alarmingly low. The World Bank in 2022 estimates that about 70% of Nigerian children cannot read and understand a simple text by age 10. Erratic and inconsistent national assessments also reveal that a large proportion of pupils in upper primary cannot perform basic arithmetic.
Thinking about the assessment issue Tobi mentioned, I realise it's a political minefield. The results are almost always bad news for politicians. Here in my state, for instance, education is tuition-free, which sounds great. But it means that if an exam shows a huge number of students failing at a particular school, the principal is in for a world of pain. To escape the government's rage, teachers and principals are essentially forced to invent better grades for their students.
When Tobi mentioned the 40,000 sacked teachers, I laughed out loud here in the office. Sadly, most states are not sincere about revealing the true staff strength of teachers on their payroll. As someone working in the primary education sector (SUBEB), it is hard to believe that I do not even know the total number of primary school teachers in my own state (O**).
Equally troubling is how this same lack of transparency is extended to development partners who spend huge amounts on our education sector.
The same goes for enrolment numbers. Officers are never allowed to give accurate figures. Two sets of enrolment sheets are always created, one "political" and one "actual." It's a behemoth issue.
Anytime I want to discuss the issue of Cement in any fora, my to go reading has always been your "Konkere" article. 😂. FF and the billionaire.
Thank you for this detailed comment. It really bothers me that we are not taking the Learning crisis seriously. It's like leaving billions of dollars on the sidewalk. A study by Eric Hanushek once estimated that we can increase global GDP by up to $700 trillion just by closing the basic skills gap in education.
I am glad I tuned in for this. Huge fan. This one struck close to work.
What can AI do for Nigeria?
I would hope to answer that question. I must apologize. I am not sure I write that good.
AI is a revolutionary tech, when I say AI I am referring to machine learning neural networks. You provide training data to a neutral network with the expectation that when your model processes data out in the wild, it can correctly provide a relationship based on training data.
This neural networks have been successful in applications such as computer vision, which I have been privileged to work with.
These neural networks has found successful applications also in Large Language Models. So when folks talk about AI these days, most likely refer to LLMs.
However, LLMs aren't going to be impactful such as the Internet or smartphone. At least not yet, there's a handful of challenges bedeviling its potential. Neural networks require incredibly high powered hardware for training and inference. A barrier to scaling to ubiquitous devices such as smartphones and laptops. There are numerous other challenges that I would restrain from listing as I suspect being verbose would obfuscate my point.
AI is next level tech built upon existing infrastructure. Infrastructure that requires high capacity educated population.
AI is already being used in dumber forms, software apps, search engines, etc.
There's nothing new to see if Nigeria hasn't utilized the existing possibilities that tech offers, and it's not really hard to see.
AI is not cheap. It's highly subsidized right now thanks to the hype. The pressure to find the successful utility for AI that would contribute to productivity is still high in the wind.
There are also the hard realities of AI use. We are witnessing in real time in the software space how speed of churning out code doesn't improve existing software products, or help you figure out stuff that people would buy.
I fear that AI compressing the training of teachers would not impact learning outcomes for students.
There's just so much hype right now that it's almost career suicide to correctly assess what is possible with AI, you jump on the hype train or you keep mum. My spicy opinion is there's a plateau right now and there's nothing new to see unless a new breakthrough in the hardware or new methodology arises. The next 1 year would be totally interesting in the AI space.
Sorry about the lengthy text. Cheers.
Interesting episode. Thank you for recording and posting it.