Artificial intelligence (AI) has long attracted an air of mystique. Depending on where you stand, it will either enrich us or kill us. Yet, rather than indulging in science fiction, we have to understand how AI affects the here and now.

If managed without caution, the rise of AI could be an unprecedented driver of inequality. But if managed correctly, it could prove to be one of the great social levellers. It could allow many in the developing world to leapfrog traditional developmental stages to become truly global, digitally enabled citizens in their own right.
AI is predicted to add US$15.7 trillion to the global economy by 2030. To ensure this benefit extends beyond the United States and China primarily, we have to make sure that more voices from the Global South are present.

So how could AI drive inequality? Historically, when technologies are distributed unevenly, exploitation and even domination can ensue. The entangled rise of gunpowder and colonialism might be one such example. As author Yuval Harari asked in an interview last May, “Will the rest of the world just become algorithmic data colonies for AI-dominating countries?” One might see the Cambridge Analytica debacle as evidence of Yuval’s fear.

This may be the doomsday fear. While this technology poses risks, we mustn’t ignore the opportunities it presents too. AI can be a transformative tool for wealth creation and public service delivery in the developing world.

Many have compared the discovery of AI to Prometheus stealing fire from the gods. Just as fire can be used for cooking or arson, depending on whose hands it ends up in, so AI remains as “good” or as “bad” as the people who use it.

In my country of Bangladesh, we have seen that, in the right hands, AI can both protect and transform lives. These successes can, and should, be replicated across the world.

For example, at the start of the Covid-19 pandemic, Bangladesh had just one RT-PCR laboratory capable of the reverse transcription polymerase chain reaction technique used to detect the coronavirus. In a country of 175 million people, this was no small problem.

But by repurposing the country’s toll-free national helpline, we were able to get citizens to self-report their symptoms. By feeding this information into an AI algorithm, we were able to successfully track and manage the disease effectively for the first four months.

This instance was instructive. While many other countries were far better resourced in terms of PCR labs and technology budgets, we were able to take a “frugal innovation” approach to AI to leapfrog the need for various other technologies and track the disease far more effectively than many of our Western counterparts. This just proves the potential of AI in developing countries like Bangladesh.

Beyond Covid-19, the impact on healthcare can be revolutionary. For example, in Bangladesh, a country with five doctors per 10,000 people, we have seen how AI can be used to dramatically reduce doctors’ paperwork load.

One remarkable example is a pregnancy monitoring app that ensures the safety and well-being of expectant mothers and their newborns by keeping all information related to illnesses and treatment online, in one location.

Similarly for education, we have been able to use AI to provide tailor-made learning and summative assessment for individual students. By automating the assessment generation process, the Noipunno initiative has produced over 2 million digital report cards, significantly easing the burden on teachers and promoting transparency in education.

With over 500,000 teachers and nearly 5 million students on board, Noipunno supports real-time monitoring, tracks attendance and ensures student development, while taking some pressure off teachers.

Yet one of the most exciting AI developments we have seen is in improving digital literacy and narrowing the digital divide. In today’s online world, digital literacy equates to power. The power to own land, run businesses, send money abroad, communicate, engage in politics and receive public services.

From India to China, how deepfakes are reshaping Asia politics

Yet developing the skills in digital literacy, especially for those in rural areas, can be extremely difficult. Here we see that AI provides another means by which we can leapfrog even the need for traditional digital literacy skills.

That is because an increasing amount of AI tools let people speak directly into computer systems, which eliminates the need for, say, word processing skills. For example, we have shown that AI can be used to help rural populations apply for land registry or identification documents by simply speaking into a phone, instead of filling out reams of online forms.

This marks a real opportunity. We are seeing that AI can provide a meaningful route for previously marginalised voices to be heard and digitally included. Yet this development is in its nascent phase.

For these innovations to be magnified and replicated across the world, we need to make sure the voices of the developing world can be heard. We must remember that AI programs remain prohibitively capital- and energy-intensive to train. For the true benefit of AI to be felt in countries like Bangladesh, these technologies should be transferred, instead of hoarded by a handful of private companies.
That’s why I was pleased that the UN’s “Governing AI For Humanity” report has spotlighted the voice of the Global South. Digital literacy has traditionally been a driver of power asymmetry. Today, we are seeing that an emphasis on AI literacy could reverse that tide. Yet that will only happen if the voices of the marginalised Global South have a seat at the table.