What Keeps Rwanda’s Women Farmers From Using AI to Protect Their Banana Crops?

Researchers examined the experiences of farmers using Tumaini, an application that identifies banana diseases and pests from plant photographs and provides management advice. The findings link women’s lower participation to unequal access to devices, household resources and practical support.

A sick banana plant can threaten the food and income a farming family depends on, making access to reliable advice a practical concern. Smartphone tools can help farmers recognise disease from photographs and decide what action to take. For many women growing bananas in Rwanda, getting hold of a suitable phone presents a bigger obstacle than learning to use one.

Research published in Frontiers in Sustainable Food Systems explores this problem in "Access, not ability: gender disparities in access to smartphone-based AI advisory tools for banana disease management among smallholder farmers in Rwanda." Jean Paul Ndayizeye and colleagues examined the experiences of farmers using Tumaini, an application that identifies banana diseases and pests from plant photographs and provides management advice. The findings link women's lower participation to unequal access to devices, household resources and practical support.

Banana bunchy top disease makes early detection especially valuable because infected plants cannot be cured, and severe outbreaks can destroy an entire harvest. Farmers need to recognise infection and remove affected plants promptly, protecting nearby crops from further spread. An application's usefulness depends on whether the person caring for those crops can reach it, understand its guidance and act on the information.

What the researchers learned from farmers

The study covered 357 smallholder farmers in Gasabo, Karongi and Rusizi, three districts with different histories of banana bunchy top disease. Researchers combined a household survey with 12 focus groups, interviews with 61 trained farmers and a year of WhatsApp monitoring involving 40 smartphone owners. The survey explored conditions before Tumaini training, and the smaller follow-up groups provided information about actual use and continued engagement.

Around 95% of surveyed farmers did not know smartphones could support crop disease management, and none reported prior knowledge of specialised smartphone disease tools. Interest was strong once the possibility was explained, with 85.4% expressing willingness to adopt such tools for better farm management, disease control, information and skills. Farmers described obstacles ranging from cost and limited digital experience to disability, difficult field conditions and competing demands on their time.

A phone in the household does not guarantee access

Smartphone ownership reached about 24% among men and 6.5% among women, despite broadly similar levels of overall mobile phone ownership. Many women owned basic keypad phones suited to calls and messages, which could not run Tumaini. Internet access showed another sharp divide, reported by 20.3% of men and 4.7% of women.

Men had roughly 4.5 times the odds of owning a smartphone after researchers accounted for education, income, electricity and other measured factors. Education and electricity access were associated with ownership, and roughly one-third of farmers lacked their own home power source. Age, income, land size and district did not independently predict ownership in that analysis, a finding that does not erase the financial pressures farmers described.

Household decisions helped explain what those pressures looked like in everyday life, with married women recording an average decision-making score of about 47% against 65% for married men. The gap was widest in Gasabo near Kigali, where men reported greater authority without a matching rise in women's decision-making power. Participants described husbands receiving priority for phones, concerns about marital suspicion, and purchases being postponed to cover school fees or other household needs.

Phone ownership brought continuing expenses for data, charging, repairs and travel to technicians, making a smartphone difficult to justify for some families. The median purchase price among smartphone owners was 90,000 Rwandan francs, and government distribution accounted for more than half of their devices. Differences in camera quality and second-hand ownership between women and men were small and inconclusive, placing the clearest divide at access to a smartphone itself.

Skills and interest did not translate into regular use

Among smartphone owners, the researchers found no clear overall gender difference in self-reported ability to use common phone functions. Calling and taking photographs were more familiar than searching online or installing applications, with some local differences such as greater social media sharing among men in Rusizi. These findings challenge assumptions about women's technical ability, though the small number of women smartphone owners and reliance on self-reports limit firm conclusions.

Language created a separate difficulty because Tumaini offered English, French and Kiswahili interfaces during the study, leaving out the participants' shared language, Kinyarwanda. Most participants could read and write Kinyarwanda, and literacy in the supported languages was much lower, particularly among women. Farmers described memorising symbols to navigate unfamiliar interfaces, showing how design choices can make an otherwise useful tool harder to use independently.

Only 24 of the 61 interviewed farmers reported using Tumaini after training, representing 39.3% of that trained group. Use reached 26.3% among women and 60.9% among men, with women describing unavailable, damaged or stolen phones, navigation difficulties and trouble understanding results. Men more often mentioned internet costs and installation problems, and many women relied on children or other people for help.

The separate WhatsApp group included 16 women and 24 men, all of whom participated during at least 10 months of monitoring. Women averaged about 35 active days across the year and men averaged 56, with comparable engagement on the days they participated. Photographs were often collected during daytime work and uploaded in the evening, connecting participation with daily schedules and opportunities to get online.

Making useful advice easier to reach

The researchers recommend access to smartphones through women's groups, continued in-person support and community discussions about who controls technology purchases and use. Kinyarwanda interfaces, voice guidance, pictures and testing on lower-quality devices could make tools more accessible to farmers with different reading skills and phones. Involving farmers in design, preparing young farmers to train others and maintaining feedback between researchers, developers and agricultural services would help support continued use.

Farmers themselves requested more training, Kinyarwanda translation and access to smartphones, with additional training mentioned by 79% of interviewees. Women who shared Tumaini knowledge reached 337 community members, including 193 women, showing the value of their local networks. Supporting those networks could spread advice to people who rarely receive it through formal services.

The study identifies relationships between access and use rather than proving that any single barrier caused lower adoption, and its selected districts and smaller follow-up groups limit broader conclusions. It did not directly measure improvements in harvests or reductions in disease from using Tumaini. Its practical message is that training needs dependable devices, understandable guidance and room for women to make decisions if digital crop advice is to become part of everyday farming.