The great education convergence

How artificial intelligence can help the poorest close the education gap

Vietnam is poorer than South Africa, yet its children learn a lot more at school. While South African kids, on average, score 343 on a harmonised test, Vietnamese kids score 519. (300 is minimal attainment; 625 denotes advanced attainment.)

The graph plots the relationship between income per capita and education quality. Richer countries have better test scores, but there is some variation around the trend; in some places, like Vietnam, kids outperform their income level. In others, like South Africa, our education is much worse than our income suggests.

That education matters in explaining the wealth and poverty of nations is not a new idea; in fact, education is often seen as the elixir of economic growth. Invest in education, the argument goes, and a country will prosper. Many African countries, for example, invested heavily in primary education after becoming independent in the 1960s. Education, many thought, would inevitably bring economic freedom.

This idea was incorporated into formal economic theory too. Paul Romer, in the early 1990s, posited that knowledge, a non-rivalrous input, is central to production; investment in education and training, therefore, would bolster innovation and drive long-term economic prosperity.

Yet while some developing countries were able to catch up over the last fifty years, others have failed to do so. In some countries, massive investments in education have failed to deliver the expected returns. Consider another two countries on the same graph above: Zimbabwe and Tunisia. Kids in Zimbabwe today perform better on a range of harmonised tests than kids in Tunisia. But Tunisia is five times richer than Zimbabwe. (The example could be more extreme: Saudi Arabia has the same education quality as Zimbabwe but is 20 times more affluent.)

In a new working paper, economists Alexis Le Nestour, Laura Moscoviz and Justin Sandefur reveal something surprising: the difference in school performance today between Vietnam and South Africa is not a new phenomenon. Many experts ascribe, for example, Vietnam’s education success to a recent period of education liberalisation. But that would be false, argues the authors. It turns out that Vietnamese kids in school in the 1960s already outperformed South African kids, despite Vietnam suffering from one of the largest Cold Wars, from 1955 to 1975. School quality, it seems, is highly persistent.

What, then, explains Vietnam’s outperformance? In a fascinating new paper, published in the Journal of Comparative Economics, economists Tien Manh Vu and Hiroyuki Yamadashow show that Vietnam’s system of imperial examinations, founded in the 11th century, might be one reason for this. They find a persistent positive relationship between the historical presence of test takers who passed the 1075–1919 imperial examinations (what they call ‘imperial elites’) and higher educational outcomes today, evident in both school attendance and standardized test scores. They suggest that this enduring impact is facilitated by the substantial returns on educational attainment, the wage premium for academic degrees, and the villages’ relative autonomy from central governance.

That education is so persistent – a topic I also discussed more than a decade ago in this post – suggests that economists have overemphasized the significance of education quality for economic growth. Although there is clearly a positive correlation, quality education may be a consequence rather than a cause of growth. It is economic growth, I would argue, that increase the returns on education, incentivising kids (and their parents) to invest in education.

That education is persistent is, perhaps surprisingly, an optimistic message for a country like South Africa, with its dreadful education test scores. (I wrote about South Africa’s recent test results, which confirm our poor performance, here.) For one, it implies that economic prosperity can be achieved through various channels, and addressing systemic issues beyond the education sector – such as improving governance, enhancing economic stability, and promoting innovation and entrepreneurship – might yield substantial progress and eventually lead to improvements in education outcomes as well.

But it also implies that the traditional education scaffolding, even with our best efforts, falls short of achieving rapid progress in education outcomes. Instead, it requires innovative approaches and tools, such as artificial intelligence, to reshape learning environments, personalise education, and directly confront the unique challenges that students face.

Although it is early still, there are already several studies that show the equalising effects of technologies such as ChatGPT. One study, using law school students, found that assistance from ChatGPT significantly enhanced student performance. Significantly, it was the weaker students that benefited most:

We also found that GPT-4’s impact depended heavily on the student’s starting skill level; students at the bottom of the class saw huge performance gains with AI assistance, while students at the top of the class saw performance declines. This suggests that AI may have an equalizing effect on the legal profession, mitigating inequalities between elite and nonelite lawyers.

Another study tested the impact of ChatGPT on academic advising. They conclude that ‘AI-powered tools, such as ChatGPT, may complement but not necessarily replace human academic advisers’.

These tools may very well serve to promote educational equity by empowering individuals from a wide range of backgrounds with the means to initiate effective methods of seeking academic advice.

Artificial intelligence will not only equalise the classroom but also the workplace. In a new paper, Ethan Mollick and a team of scientists set up a randomised control trial using ChatGPT in the Boston Consulting Group. Here are their results:

Across a set of 18 tasks designed to test a range of business skills – from analysis to idea generation to persuasion – consultants who had previously tested in the lower half of the group increased the quality of their outputs by 43% with AI help while the top half only gained 17%. Where previously the gap between the average performances of top and bottom performers was 22%, it shrunk to a mere 4% once the consultants used GPT-4.

It is exciting that there are already several world-class #edtech start-ups in South Africa – Mindjoy, FoondaMate, Rekindle Learning – that combine the power of artificial intelligence with the need to tackle South Africa’s woeful education quality. What these companies now need, on the supply side, are risk-tolerant investors and a supportive institutional environment that removes unnecessary regulatory constraints. The recently launched Edu Invest initiative by Wesgro and the Western Cape Education Department is an excellent example of such a supportive environment.

But supply-side access is only part of the solution. We need to make sure that not only is education available, but that kids and their parents really want it. To do this, we need a government that removes the hurdles that inhibit economic growth. A strong economy that creates jobs will encourage families to put money into education. This is especially true now that AI can help make education even better, particularly for those who need it the most.

An edited version of this article was first published on News24. Image created with Midjourney v5.2.