AI sparks a new agricultural revolution in the Global South

On a sunny April morning in Nairobi, the United Nations compound buzzed with more than diplomatic chatter. Instead, the air crackled with ideas that could shape the future of farming in the Global South. Scientists, software engineers, and agricultural innovators from across the globe gathered for CGIAR’s Science Week side event titled “AI-Powered Innovation: Accelerating Research for Agri-Food System Transformation.” Their shared mission? Harness artificial intelligence to revolutionize food systems under mounting pressure from climate change, resource scarcity, and hunger.

This was no ordinary tech conference. Attended by over 250 participants in person and virtually, the event was a four-hour deep dive into the very real and rapidly evolving role of AI in transforming agriculture—from seed to satellite, and from chatbot to crop field.

AI on the farm: from flood forecasts to voice-powered advice

The event opened with a powerful vision. Shalini Gakhar, data scientist at the International Rice Research Institute (IRRI), painted a vivid picture of a future where drones, sensors, and voice-based AI guide farmers through climate shocks and planting decisions. “This future is already here,” she said, “but the challenge is not technology—it’s trust, readiness, and responsible deployment.”

That future was echoed in a keynote by Aisha Walcott-Bryant, Head of Google Research Africa, who showcased AI tools already saving lives: real-time flood forecasts deployed in 90 countries, wildfire alerts, and hyper-local weather predictions available through Google Search. Her colleague, Abigail Anka, Research Software Engineer, Google AI, demonstrated how a building-mapping tool evolved into a powerful engine for detecting agricultural field boundaries—laying the foundation for smarter, more responsive food systems.

Generative AI and the power of local voices

In one of the most engaging sessions, CIMMYT’s Monitoring, Evaluation and Learning Manager, ICT for Agriculture, Andrea Monsalve, and ICT for Development Specialist, Satish Nagaraji explained how generative AI tools are being localized to serve farmers who often lack literacy, internet access, or trust in conventional digital platforms.

In Bihar, Kenya, and Mexico, farmers now ask questions on WhatsApp in local dialects and receive personalized audio advice from AI-trained systems. Some even use AI-powered phone calls to contribute data through interactive voice response systems. The systems are not static, feedback from farmers loops back into the AI models, making them smarter and more relevant with every interaction.
“It’s not just about pushing information,” said Monsalve. “It’s about co-creating solutions with farmers.”

Listening to farmers with AI, literally

That theme of listening took on literal meaning in the Artemis panel, which introduced SIKIA, a Swahili word meaning “listen.” This AI tool captures farmers’ open-ended voice responses and uses speech recognition and natural language processing to extract insights. In one trial in Tanzania, local youth recorded responses from 480 farmers about common bean varieties. The AI then analyzed these to reveal preferences – like cooking time and pod maturity – that often go unmeasured by scientists.
“You’d be surprised,” said panelist Violet Lasdun, PhD Student, Alliance Bioversity-CIAT. “Women consistently prioritize taste, while men emphasize plant height and yield. And women get to the point faster.”

The Artemis project, launched as a global public good, also includes low-cost phenotyping carts (called “Bruno”) and annotation teams to train AI on crop traits—making sophisticated data collection possible even in the most remote farms.

AIEP: Designing for the underserved

Another standout initiative was the Agricultural Information Exchange Platform (AIEP), co-funded by the Gates Foundation. Through collaborations with organizations like Digital Green and Viamo, the platform delivers AI-powered advisory tools to farmers via SMS, IVR, and WhatsApp, designed specifically for those with low literacy or digital access.

Christian Merz from GIZ, a lead partner, highlighted a user named Rose—a Kenyan farmer using a basic mobile phone to query AI in her own dialect and receive tailored responses. “This is not science fiction,” he said. “This is happening now.”

Data from space and genes: satellites and genebanks go smart

Not all the AI work stayed grounded. The Tracking Adaptation Progress in Agricultural Systems (TAPAS) platform from the University of Galway uses satellite imagery to evaluate climate adaptation investments, tracking irrigation efficiency, soil moisture, and even methane emissions from rice fields. Meanwhile, IRRI’s genebank in the Philippines is using machine learning to speed up seed screening for climate-resilient traits, having evaluated more accessions in one season than in the previous 45 years combined.

A call to collaborate, not just innovate

Throughout the event, one message was clear: AI in agriculture must be ethical, inclusive, and collaborative. From African language datasets to low-tech phones and gender-sensitive designs, the spotlight remained firmly on contextual relevance and accessibility.

“We need AI that listens, literally and metaphorically,” said Jacob van Etten, Director, Digital Inclusion, Alliance of Bioversity & CIAT. “Because the real innovation isn’t the algorithm. It’s what happens when a farmer’s voice shapes the science.”

As the event closed, the energy didn’t dissipate, it deepened. A sense of shared purpose filled the halls: AI could indeed reshape agriculture in the Global South, but only if built with, and for, the people at its heart.