Bridging the Gap: AI’s Potential in Africa’s Data Desert – Sponsor Content

The First Industrial Revolution was fueled by steam and water power. The Second relied on electricity, the Third on electronics and information technology. The Fourth Industrial Revolution, says Klaus Schwab, founder of the World Economic Forum, is unfolding as “a fusion of technologies”—the Internet of Things, the cloud, and AI, among others.

Data is the engine of the Fourth Industrial Revolution. Every one of us continuously generates data—economic, social, political, and medical, among others. “Data is a seed for economic transformation,” says Kate Kallot, founder and CEO of Amini, a Nairobi-based environmental-data company. Data is at the center of international markets and shapes entire sectors.

But where data can’t be collected, read, or analyzed, people, regions, and economies—even weather—can be effectively “invisible,” says Kallot. By this definition, Africa is “the most data-scarce continent,” she says. Over the next decade, Africa’s working-age population will increase by 450 million—nearly 70%—and make its mark on the world. Yet broad swaths of the continent are not represented, economically or politically, in the tools we use more and more to understand the world, because most of the sub-Saharan region lacks broad-based access to the high-speed internet.

As this new infrastructure blossoms in sub-Saharan Africa alongside an eager, youthful workforce, bold innovators are harnessing AI in surprising ways.

But as the sub-Saharan population is exploding, so too is excitement about AI. According to a recent study from Google and Ipsos, the Global South expresses more optimism about AI than any other world region—in fact, says Miishe Addy, founder and CEO of AI-driven logistics and financing start-up Jetstream Africa, “I’ve gotten to know more than a dozen diasporans who have been educated and built networks outside who have come back to Africa to build businesses.” In countries where survey respondents were most likely to think AI will change most jobs and industries in the next 5 years, they also tended to be the most positive about that change. In South Africa, for example, 59 percent declared that changes in jobs and industries as a result of AI in the next five years are “probably a good thing.”

In the years to come, readiness for the AI revolution will be crucial, largely because emerging markets may have the most to gain. In a UN survey of AI readiness in nine global regions, sub-Saharan Africa ranked last. The necessary expansion of technological infrastructure, along with training and upskilling, will bring people online, giving them access to information and opportunities, and empowering them to share with the world—using images, text, or videos—what they think, feel, and want. The proliferation of both infrastructure and new users will in turn up-level AI models and enable further growth and innovation.

In many ways, this process has already begun. As this new infrastructure blossoms in sub-Saharan Africa alongside an eager, youthful workforce, bold innovators are harnessing AI in surprising ways to address challenges and harness potential in three key areas: agribusiness, finance, and transportation.

I. AGRIBUSINESS

Sub-Saharan nations are among those hardest hit by global warming, with populations that depend largely on agribusiness (as 52% of employed people do) experiencing worsening floods and droughts. Yet Africa—with only one-eighth of the minimum density the World Meteorological Association requires for weather stations—has a scarcity of both historical and real-time data that profoundly impairs meteorological predictions.

Using AI-powered weather forecasting, researchers are working to close this gap. Companies like Amini in Nairobi and projects like Google’s AI for Weather Forecasting in Accra, Ghana use real-time satellite data and historical data to develop highly accurate forecasts that can help the broader population, and farmers in particular, cope with increasingly erratic weather.

All of these models are trained in the United States or Europe. Therefore, they will get things wrong in Africa.Rob Floyd director of innovation and digital policy, ACET

This data is critical for food security. “For a farmer,” says Emmanuel Asiedu Brempong, a research engineer on Google’s AI for Weather Forecasting team, “knowing the weather is like a magic wand.” Rainfall predictions can help determine which crops to plant (for instance, rice requires more rain; tomatoes less); and when to administer soil treatments (rainfall helps to dissolve and disperse grains of fertilizer, for instance, but washes away liquid pesticides).

But agribusiness demonstrates how data scarcity can lead to data bias. One bovine-health app found that virtually every Holstein cow it evaluated in Africa was malnourished, says Rob Floyd of the African Center for Economic Transformation (ACET). In fact, Floyd points out, the cows were perfectly healthy. The AI needed to be trained, using local data, that African cattle are naturally skinnier. “All of these models are trained in the United States or Europe,” Floyd observes. “Therefore, they will get things wrong in Africa.”

II. FINANCE

People without bank accounts may struggle to prove their creditworthiness. But they do have payment and usage histories, along with networks of contacts. AI technologies can construct an identity from these patterns and connections to create a picture of a person’s finances. “That gives someone, frequently for the first time, a document where they exist as a consumer,” says Davide Strusani, an economist and consultant who has studied AI in emerging markets. “The moment you have a digital identity, you can ask for a digital bank account, then a microloan, and maybe a credit card.”

In Uganda, Yabx aggregates such variables as utility payments, network-usage patterns, and mobile wallet behaviors to assess creditworthiness. The automatic disbursement of payouts via non-bank payments providers, like M-Pesa—active in seven countries, including Kenya, Tanzania, and Lesotho—reduces the barriers farmers without bank accounts may encounter when they seek to purchase financial products related to the climate crisis, such as crop insurance.

Customers register for M-Pesa at authorized outlets—often mobile-phone stores or retailers like barbers, butchers, or bakers—where they exchange cash for electronic funds, which they can use to purchase insurance policies. In Kenya, for example, M-Pesa processes premiums and payouts for a microinsurance provider, ACRE Africa. ACRE collects premiums prior to the start of the season and compensates farmers roughly three to 20 weeks following the harvest. The policy covers crop losses during four distinct chronological stages—germination, vegetative, flowering, and maturity—and the losses in each stage are calculated at the end of the season to determine the final payout. “If people start texting money,” Strusani says, “they start trusting the system.”

Making affordable insurance accessible to all shores up a healthy marketplace; underinsurance can cause widespread economic shock.

A Lloyd’s/Centre for Economic and Business Research report notes that emerging markets represent $160 billion, or 96 percent, of the total global insurance-protection gap, and that “identifying barriers to the scalability of insurance solutions,” particularly in the area of weather risk, will be critical to addressing this gap. Making affordable insurance accessible to all shores up a healthy marketplace; underinsurance can cause widespread economic shock.

African regulators recognize the importance of privacy around financial transactions. Due to the sensitivity and stakes of these issues, governments tend to err on the side of caution. This can be a blunt but necessary approach to a delicate operation—the shepherding of the expansion and integration of innovative technologies—and it can have the inadvertent effect of hindering innovation. Recently, for example, concerns about transactions potentially related to money laundering led Nigeria to pause the domestic expansion of payment-service providers.

III. TRANSPORTATION

The demand for transportation in sub-Saharan Africa is increasing rapidly due to population growth, urbanization, and, in some regions, rising prosperity—all of which leads to more traffic and higher volumes of freight. AI technology holds promise for making transport safer, more reliable, more efficient, and cleaner. It has the potential to attract private-sector investment, which in turn creates markets and allows countries to reach more underserved populations.

In West Africa, start-up Jetstream Africa—the continent’s largest digital freight forwarder—aims to serve the continent’s 2 million import and export businesses, offering features like AI-powered process automation and trade finance for underserved small and midsize businesses. By 2026, the company’s AI platform will use historical performance data to monitor supply-chain reliability—to predict delivery and loan repayment times. “Jetstream uses AI because it makes our work easier, and cheaper for our customers,” says Addy. “I think you’re going to see more applications and more specific use cases of AI to expand access to services that our customers would not otherwise be able to afford.”

The Challenges

Data scarcity is one of several daunting obstacles to the growth of AI-driven technology in Africa. There are also technological, regulatory, and political challenges, all made more complex by the fact that Africa comprises 54 different nations. “The African Union is a very large, very slow bureaucracy,” Floyd says. Unifying its various policies—which in the overwhelming majority of cases have yet to be written—is a complex undertaking.

1. Physical infrastructure

Of the challenges facing the development of AI on the African continent, the most pressing may be the lack of infrastructure—both technological and regulatory. “Internet connectivity in Africa happens mostly in urban areas, and often only on low-speed networks,” says Strusani.

But he notes that the recent installation of two new submarine cables will mean a massive increase in capacity. More broadly, AI itself can assist in the building of high-speed networks. Analysis of satellite imagery by AI-driven projects like WorldPop and Google’s Open Buildings offers invaluable information on population distribution and density that can guide the installation of both technological and electrical infrastructure.

2. Regulatory infrastructure

To date, only one African nation, Rwanda, has announced a national AI policy. Whereas companies or governments drive innovation in many countries, Kallot says, in Africa, “it’s actually a bottoms-up innovation driven out of grassroots communities. And the people who are driving it are not usually the people who have a seat at the table when it comes to discussing regulations.” There is a sense that young AI entrepreneurs are neither represented nor understood by the governing and policymaking generation. “The government too often regulates what it doesn’t understand,” says Addy.

3. The data-sovereignty conundrum

National policies on cross-border data-sharing are “a huge issue,” says Strusani. Having seen foreign companies collect and take ownership of African data, some African countries have implemented policies restricting the flow of data across national borders or requiring domestic data mirrors.

This struggle for data sovereignty has far-reaching implications. For instance, why would an entrepreneur seeking to innovate set up shop in a country with a restrictive policy? In a data vacuum, how can regional policymakers working on issues of governance make decisions? What is the sense of building expensive new data centers that create minimal employment, harm the climate, and are redundant with the cloud?

Such policies create a lack of cohesion within the African Union and can stifle the growth of smaller economies. “Big countries like Nigeria, South Africa, and Kenya don’t necessarily need to trade with Benin or Togo or little countries,” says Floyd. “They have less interest in regional integration.”

4. The problem of buy-in

Achieving buy-in from governments is critical, and one argument for doing so is that AI can create a healthier body politic.

Data generated by Google’s Open Buildings project can help us understand the impact of natural disasters, in order to guide emergency response. Such information can also assist with everything from planning vaccinations to registering changes in urbanization, says Abigail Annkah of Open Buildings.

Another initiative, WorldPop, uses satellite imagery and Google’s Open Buildings Dataset to assemble granular population figures for the many low- and middle-income countries whose registers may be outdated or incomplete (to cite one example, the Democratic Republic of Congo—a nation the size of western Europe—took its last census in 1984). Population data can guide everything from the allocation of resources to the planning of schools to the positioning of healthcare. “It underlies everything about equitable, efficient, and effective governance in general,” says Andy Tatem, director of WorldPop.

To achieve buy-in for its various initiatives, WorldPop has built critical relationships of trust with governments and UN agencies, and co-develops solutions with them—in effect, serving an ambassadorial function for AI that can open the same doors to private industry.

AI-powered applications could increase Africa’s economic growth—helping to end poverty and boosting shared prosperity—by $2.9 trillion by 2030

UNIFYING AFRICA AROUND AI

Recently, there have been several promising developments in building a unified AI policy for the 54 nations in the African Union—including the first UNESCO-Southern Africa sub-Regional Forum on Artificial Intelligence, which made recommendations in areas like AI governance, environmental protection, and gender inclusion. In March, the African Union Development Agency (AUDA/NEPAD) released a white paper on the regulation and responsible adoption of AI.

AI is helping to unify the nations of Africa on a grassroots level as well. Google’s 1,000 Languages Initiative—a commitment to build AI models that will support the world’s 1,000 most widely spoken languages—breaks down barriers in communication to connect people and help them better understand the world—and each other. Its recent expansion—a joint effort by Google Research, Google DeepMind, and Speech teams—is its largest ever, and includes 110 languages. Some have over 100 million speakers; others are spoken by small communities of Indigenous people; and a few have almost no native speakers but are the subjects of active revitalization efforts. Twenty-two are African languages, among them Ga (745,000 speakers in Ghana), Tshivenda (1.3 million in South Africa, Zimbabwe, and Mozambique), and Wolof (8 million in Senegal).

One way to conceptualize the way forward is through the lens of Google’s 2020 Sprinters report, which offers a model for harnessing AI to accelerate economic development in emerging markets. The report focuses on readiness in four critical areas—infrastructure, skills development, technological innovation, and the advancement of policies to foster a thriving AI ecosystem.

The GSMA, an international nonprofit representing mobile-network operators, recently estimated that AI-powered applications could increase Africa’s economic growth—helping to end poverty and boosting shared prosperity—by $2.9 trillion by 2030. Private-sector solutions will play a critical role here; but widespread access to computing and the internet, made possible by new infrastructure in the public sector, will provide workers with vital ground-floor entry into the digital market. Once realized, the benefits of AI could swiftly and dramatically transform local economies. “New technologies invented in the early 20th century took more than 50 years to reach most countries,” The Economist recently noted. But AI will not take nearly as long, the magazine predicted—because it will reach people directly, via an object “so many already possess: the phone in their pockets.”

Originally Appeared Here