Can AI Help Feed the World? UVA Team Takes on Climate-Driven Food Crises in Lesotho

Allison Barrett Carter
Image
fields tilled in Lesotho South Africa

A new UVA Environmental Institute Climate Collaborative team is working with farmers in Southern Africa to develop a customized AI tool they believe can improve yields.

University of Virginia professors Molly Lipscomb and Terence Johnson both believe that researchers have a responsibility to tackle challenges not just their own backyard, but also those that span the globe. This has inspired them to join forces to see if artificial intelligence can help solve critical food security problems for smallholder farmers in Lesotho, South Africa. 

Lipscomb, associate professor in the Frank Batten School of Leadership and Public Policy and Johnson, assistant professor in the UVA School of Data Science, are leading the effort alongside scientists from the National University of Lesotho and community builders at 4D Climate Solutions to develop solutions for farmers in Lesotho.

Image
farmers in the fields in Lesotho

Farmers in the Lesotho area of Southern Africa are grappling with volatile conditions as climate change affects growing seasons. (Photo contributed.)

“Food security is such an obviously important problem, and it is getting worse with climate change,” Lipscomb explained. “We have a responsibility to do something, especially in terms of these smallholder farmers. These farmers are already barely subsistence farmers, and the increased volatility in weather that comes with climate change has particularly severe welfare consequences for them.”

Lipscomb, Johnson, and the team are developing an artificial intelligence (AI) tool they hope will help farmers maximize yields and increase growth. Recently announced as a new UVA Environmental Institute Climate Collaborative, the interdisciplinary team works with community members to help farmers in the Central Region of Lesotho make informed decisions and adopt effective climate-adaptation practices.

The government has published documents and recommended procedures detailing best practices for the farmers. If utilized by the community, these documents contain scientifically backed tips specific to their location, which can help navigate changes in growing season, soil acidities, extreme temperatures, droughts, and more.

The problem: the documents are hard to reference and, most of the time, the information doesn’t reach the farmers in time to be helpful.

The team is seeing if AI is up to the task. Can they code an AI tool specifically for farmers in Lesotho to provide them with the quick, easy-to-access information needed to be productive growers in the face of rapidly changing environmental conditions?

Image
the team stands smiling

Community members, Lesotho extension workers, and multiple organizations have teamed up to see if AI can offer easy-to-access solutions for farmers in Lesotho. (Photo contributed.)

"There are these great resources compiled by agricultural experts in Lesotho, but few farmers have genuine access to them,” shared Johnson. “They can be technical in ways that require clarification, or organized in complex ways, or it's not obvious what information you even want to retrieve. By reshaping these complex, multi-modal materials and making them accessible to farmers through an AI co-piloted by a trained agricultural extension worker, we can make these resources more accessible for the farmers that need them, boosting productivity in a responsible way that draws on the entire knowledge base."

There are a few challenges to the AI tool that Lipscomb, Johnson, and team members are addressing. First, and most importantly, bad AI results could be more harmful than having no help at all. The researchers are diligent in their work, ensuring this new AI tool is accurate. This is also why Lipscomb is convinced that a specialized and customized AI tool is needed, rather than relying on current AI tools more generally available, which provide results sourced from across the entire Internet rather than focusing on farming issues specific to Lesotho. 

Second, access to the AI tool could be difficult in rural areas of South Africa, such as Lesotho. Through the partnership with 4D Climate Solutions, the researchers have extension workers who visit and work alongside farmers to access, train, and refine the AI tool.

“Around the world, people are coming up with technologies to increase productivity and yields. Food security problems feel solvable,” Lipscomb said. “Yet we still seem so far from solving agricultural problems such as the ones being experienced in Lesotho. I feel a responsibility to do what I can.”

The researchers are in the early stages of the project, hoping to finalize the AI tool over the summer of 2026. The Lesotho community has been helpful and involved as the project develops. For the UVA professors in Charlottesville, Virginia, the work is mission-central to what they do, and they look forward to bringing what they can to make agriculture more productive in the region, even if it is miles away.

"We're studying a small piece of how you might use AI to help farmers, but when you aggregate all these efforts across the many nonprofits and researchers working to end poverty, it results in a more abundant and humane future,” said Johnson. “That's why universities exist: to create knowledge and turn it into progress."