Tanzanian researchers to develop ChatGPT-powered chatbot for early detection of diseases in maize farms
TANZANIA – The Bill and Melinda Gates Foundation has funded researchers at the Sokoine University of Agriculture (Sua) with US$100,000 (Sh245 million) to develop a Chat Generative Pre-Trained Transformer (ChatGPT)-powered Swahili chatbot for early detection of diseases in maize farms.
Ms. Theofrida Maginga, a Sua assistant lecturer currently pursuing a Ph.D. in Internet of Things with a specialization in Embedded Computing Systems, and a colleague from the African Centre of Excellence on Internet of Things (IoT) at the University of Rwanda, Dr. Jimmy Nsenga will spearhead the research.
According to Citizen TZ, the envisioned technology will enable smallholder farmers with limited literacy and scarce resources to quickly and easily detect crop diseases such as Northern Leaf Blight.
Reports show that 40 percent of maize yield in East Africa is lost, threatening the realization of targets to ensure food security and zero hunger amidst a drastic increase in the world population.
The researchers, therefore plan to integrate artificial intelligence (AI) with Internet of Things (IoT) technologies and use non-invasive sensors to monitor non-visual early disease indicators in a bid to prevent yield loss and spur up production.
Ms. Maginga said advanced disease cycles include different phases such as inoculation, penetration, infection, incubation, reproduction, and survival, therefore requiring higher intervention measures to control the diseases.
“We are developing sensors that will be placed next to plants with disease and those without disease. In the laboratory tests, when a disease was introduced to certain parts of the crop, some differences were noted over time,” she said.
According to the researcher, the nonvisual signs incorporated in the maize crop include volatile organic compounds (VOC), ultrasound, nitrogen, phosphorus, and potassium (NPK) uptake.
“The main goal was to come up with a simple device that uses low power and that requires less interaction with individual farmers, but that can still be placed in the field, and farmers can be told what is going on in the field,” she added.
Ms. Maginga said most farmers make late detections of disease outbreaks in their maize farms, triggering huge losses.
According to her, in this sense, diseases keep spreading to other field areas, broadening the loss to almost half of the farmers’ expectations. The impact is passed on to seeds, therefore giving poor results when grown in the future.
Focus on rural farmers
Particularly, Maginga highlighted that the technology will involve testing to establish ways to reach the majority of smallholder farmers in remote areas.
She elaborated that the duo will develop and integrate a Swahili chatbot to interact with farmers in their local languages in a culturally-sensitive manner and perform model validation and field testing.
According to her, what made funders chip in was the fact that technology can be integrated with chatbots, where answers could be provided to questions raised by farmers.
“It means chatbots can be asked simple questions and provide room for the generation of appropriate answers. Therefore, we are working to bring ChatGPT in terms of chatbots to farmers,” she said.
She added that having the IoT part will limit forcing farmers to report to extension officers after detecting diseases, but instead, the technology will do that, adding that the technology acts by “minimizing interaction with human beings.”
Ms. Maginga stated that before the full deployment of the project, some training will be conducted for farmers and extension officers on the best way to interact with the technology, which will be known as MKULIMAGPT.
According to her, the technology will be deployed to farmers no later than December 2023, after the testing is concluded in mid-November.
For all the latest grains industry news from Africa, the Middle East and the World, subscribe to our weekly NEWSLETTERS, follow us on LinkedIn and subscribe to our YouTube channel
