Western AI models "fail spectacularly" in farms and forests abroad

AI models from Western tech giants fail in overseas agricultural settings - Rest of World Skip to content By Rina Chandran 12 March 2026 AI models built in the West often fail to function correctly in poorer nations because they are not trained on local data. Effective use of AI in agriculture requires adaptation and local ownership. There is a risk that a focus on profit by big tech firms and large agriculture companies will hurt farmers. When scientist Catherine Nakalembe set out to map crop types in western Kenya, she had plenty of data from satellite images, but couldn’t use artificial intelligence to analyze it because the data could not recognize local crops. She decided to collect her own data, fitting GoPro cameras on the helmets of dozens of volunteers, and training the facial recognition technology to identify maize, beans, and cassava. They collected over 5 million images in two weeks. Nakalembe uses machine learning, computer vision, and deep learning models to map cropland, classify crop types, and estimate yields in Uganda, Kenya, Senegal, and other African nations. But most AI models are trained on European and U.S. data, and are largely useless unless they are adapted for local contexts, she told Rest of World . “AI systems built in the West often also fail to account for the contexts of the Global South, including high internet costs, limited bandwidth, and a lack of labeled training data,” said Nakalembe, an assistant professor at the University of Maryland, and Africa program director at NASA Harvest, which uses satellite imagery to improve agricultural production. “If these systems aren’t adapted, they remain irrelevant, potentially deepening existing inequalities in wealth and access to resources, [and] there is a risk that these systems prioritize corporate and company profit over farmers,” she said. As AI systems become more sophisticated and accessible, governments and organizations are keen to use them to tackle issues such as deforestation and food security to benefit farmers , fishers, and other rural communities. Agriculture provides livelihoods for more than 2 billion people in low and middle-income countries, and they are exposed to climate change impacts that hurt crop yields and reduce incomes. If AI assumes literacy, connectivity, or decision authority, it only benefits better resourced farmers first and widens inequality.” Technology, particularly AI, can be a solution, Oren Ahoobim, a partner at consultancy Dalberg Advisors in San Francisco, told Rest of World . While satellite imagery, video, and chats have been in use for some time, the quality and availability of the underlying data that powers these technologies “is dramatically better now, so the outputs are much better and can be better trusted,” Ahoobim said. “This translates to better information for farmers, including more predictive information , which enables farmers to make better decisions earlier on in terms of what to plant, how much fertilizer to use, how to manage disease, etc.,” he said. Ending hunger is one of the 17 sustainable development goals of the United Nations for 2030. But the goal of zero hunger is likely to be missed , with about 28% of the global population — around 2.3 billion people — “moderately or severely food insecure.” AI is being used in several ways to address this: In Brazil’s Pará state, AI turns real-time coastal data into actionable WhatsApp voice alerts for fishers and oyster farmers. Microsoft’s AI models use bioacoustics to monitor deforestation in the Amazon forest. Digital Green’s FarmerChat Android app reaches more than 1 million farmers in South Asia and Africa, using generative AI to answer queries in 16 local languages, and diagnosing crop issues from uploaded images. FarmerChat has answered over 8 million questions, co-founder and chief executive Rikin Gandhi told Rest of World . The team trained and reinforced small language models with over 120,000 farmer queries and answers developed with agronomists and veterinarians in vernacular languages, and in the manner that farmers “actually speak” — without the formal names for seeds or chemicals, Gandhi said. “Agriculture is hyperlocal: Soil type, rainfall, altitude, pests, and markets vary village to village. Model learning must stay close to those realities,” he said. Building trust is also key — the AI must support farmers and ensure equity of access without being used for credit scoring, risk profiling, or compliance. “If AI assumes literacy, connectivity, or decision authority, it only benefits better resourced farmers first and widens inequality,” Gandhi said. In India, when Farmers for Forests, a conservation group that aims to increase forest cover, tried a popular open-source model to analyze data from the western state of Maharashtra, “it failed spectacularly,” co-founder and director Arti Dhar told Rest of World . “It missed over half the trees because it was trained on North American forests,” she said. The team manually annotated drone imagery from 80 different land parcels, labelling about 55,000 individual trees. “It was a clear lesson that you cannot simply parachute Western AI into the Global South and expect it to work,” Dhar said. Dhar’s team uses drones to create 3D maps of farms, and a custom-trained AI model built on Meta’s Detectron2 to identify every tree and measure its height and canopy. With that data, the diameter of tree trunks can be estimated to calculate carbon sequestration, which can generate incomes for farmers, Dhar said. Their WhatsApp chatbot, ChutkiAI, supports farmers. An AI system can perform technically well and still fail farmers if it ignores economic and ecological realities.” “Local ownership and adaptation are critical … otherwise the promise of AI will remain concentrated in the hands of a few,” she said. The technology is also “only a piece of the puzzle. … The most accurate AI in the world is useless if it isn’t embedded in a system of trust and aligned with the real-world economic needs of farmers.” Digital tools in farming are a big business, with the market worth about $30 billion last year and forecast to nearly triple to $84 billion by 2034. Tech companies including Google , Microsoft , Amazon , IBM, and Alibaba all have AI programs to tap this growing market. Yet there is a risk that AI becomes a new form of digital colonialism, experts have warned, with big tech firms extracting data from poor communities to train a proprietary model and then selling a service or product back to them. Tech companies are already working with large agriculture firms to influence what crops are grown and how, according to the International Panel of Experts on Sustainable Food Systems, a think tank. Focusing on only the most productive and profitable crops — corn, rice, wheat, soybeans, and potatoes — can wreck local food systems and hurt farmers, it said. This is what organizations must guard against, Gandhi said. If AI optimizes only for short-term yields and ignores larger issues such as water depletion, soil degradation, or energy use, it can erode long-term resilience. “An AI system can perform technically well and still fail farmers if it ignores economic and ecological realities,” he said. “The real measure of AI in agriculture is whether it strengthens farmer agency, improves profitability, supports sustainability, and works for women and men. That depends entirely on building with farmers, not just for them.” Read more stories Tech Giants The Gulf built oil pipelines to avoid Hormuz. It's now doing the same for data Saudi Arabia, Qatar, and the UAE are financing competing data corridors through Syria, Iraq, and East Africa to bypass the two maritime choke points that threaten their digital connectivity. By Indranil Ghosh Innovation Black-box AI and cheap drones are outpacing global rules of war From 3D-printed drones to Anthropic’s Claude, advanced technologies are making conflict more accessible and less accountable — leaving human oversight at risk. By Rina Chandran Innovation The Gulf war is reshaping how Asia works The global oil crisis triggered by the U.S.-Iran war is making Vietnam, Thailand, the Philippines, Pakistan, and India get conservative with energy. By Ananya Bhattacharya