As foreign aid falters, can AI step in?

For decades, foreign aid has followed a familiar script: money flows from wealthy countries to poorer ones, funding programs meant to boost health, education, and economic opportunity for communities across the world. But as budgets shrink and tensions rise, that model is showing its age — leaving many to turn to artificial intelligence to take its place. “There are many tech companies that, in the last year or two, have committed to upskilling millions of people in India, and Southeast Asia, and parts of Africa,” said Maha Hosain Aziz, a professor at New York University and co-chair of AI policy at The Digital Economist think tank. “So from my perspective, AI and the individuals or firms using that technology are, in a way, filling the aid gap that has been with us for many, many, many years.” One example of that is Digital Green, a Kenya-based organization that uses AI to deliver personalized, climate-smart advice to farmers. By leveraging a GPT-3 language learning model, Digital Green created an AI system called Farmer Chat — one that allows farmers to ask questions about specific agricultural problems, and receive feedback on how to improve crop yields, manage pests, or adapt to shifting weather patterns in real time. With GPT-3 delivering hyper-local advice tailored to a farmer’s specific needs, it’s a sharp contrast to the often-blanket recommendations of traditional development services, explained Rikin Gandhi, Digital Green’s chief executive officer, speaking alongside Aziz at the Devex Impact House during the United Nations General Assembly. “We can think of a less dependent form of aid and development,” he added.“The farming communities that we serve are able to access models directly on their phones — and the cost of doing so is so low that they can be able to do this on a sustainable basis.” Founded in 2008, Digital Green began by sharing farming best practices through community-led video programs. Since then, it has partnered with governments and local organizations in more than a dozen countries, reaching over 6 million farmers — the majority of them women — with digital tools and training designed to boost productivity and resilience. Digital Green is far from the only organization looking to integrate AI with development. In the United States, the Grant Assistant AI platform is helping organizations apply for funding, using natural language processing to draft proposals tailored to donor priorities. In the humanitarian world, the Famine Action Mechanism leverages AI algorithms to provide early warnings for food shortages, giving governments and aid agencies a heads-up before a crisis escalates. But even with the sprawling growth of AI-infused projects, both Aziz and Gandhi warned that embedding such technologies into aid is far from foolproof. There’s a risk of swapping one type of dependency for another, Aziz said, with communities becoming reliant on AI solutions just as they once were reliant on development aid. Most language models are trained on data and information from the global north, Gandhi added, a process that may exclude the insights those in other parts of the world — including the rural farmers Digital Green focuses on — need. And as nations compete to export their AI systems — often bundled with their values and governance approaches — the technology could deepen divides rather than close them. Despite AI’s explosion, about one-third of the world’s population — roughly 2.6 billion people — remains offline, according to the International Telecommunication Union. And while the number of people using generative AI tools is rising, access to the most advanced models is still concentrated in high-income countries and large urban centers, leaving many rural communities behind. Still, Aziz argued, the scale and speed of change mean that turning back isn’t an option — and that development actors will need to adapt rather than resist. “I don’t think we can stop AI, nor can we stop the export of AI,” she said. “It’s not so straightforward, but I think it’s positive. I think we’re moving in the right direction.”