Building farmers' confidence in precision agriculture technologies is the goal of a National Science project led by Maaz Gardezi, an assistant professor of sociology and rural studies and an affiliate faculty of natural resource management at South Dakota State University.
“Precision agriculture technologies, such as artificial intelligence-based decision support systems, use localized farm data to generate site-specific recommendations on when to plant, seed, spray and harvest. Yet, some farmers lack confidence in these recommendations and have concerns with data ownership and privacy, which can impede the adoption of these precision ag tools,” Gardezi explained.
The four-year, $3 million NSF project will use a unique approach to develop, test and implement new precision agriculture tools and public policies that are socially and economically feasible for farmers, rural communities and the environment.
“Our approach is using a network of farmers, nongovernmental organizations, industry representatives and Extension agents, not just as users, but as co-designers and co-evaluators of technology,” Gardezi said. Other SDSU faculty working on the project are Distinguished Professor David Clay of the Department of Agronomy, Horticulture and Plant Science and assistant professor John McMaine of the Department of Agricultural and Biosystems Engineering.
The living laboratory approach considers the needs and concerns of those who will use and benefit from these technologies during the initial stages of product development. “For example, farmers will be evaluating these technologies and will be teaching each other and us—through in-depth interviews and experimental design— how to use these tools and how to improve them,” Gardezi explained.
At the same time, involving nonprofit organizations and industry experts in the research will help balance artificial intelligence-based innovations with societal needs and environmental demands,” he noted.
Gardezi sees this research as having “a far-reaching impact on how we approach innovation collectively and responsibly, including who will use it, who will not and what will it replace. It’s a balanced overview of how innovation should be done.”
Tackling sustainability in two states
The on-farm research will involve 48 farmers—24 in South Dakota and 24 in Vermont. To do this, he will be working with two University of Vermont faculty—professor of civil and environmental engineering Donna Rizzo and professor of public policy and computer science Asim Zia.
“This is the first time this (living laboratory) approach will be applied to agriculture in two distinct regions in the United States,” Gardezi said. Vermont has a diverse organic agricultural sector as well as dairy farming, while the South Dakota cropping systems tends to focus on corn and soybeans.
Though the SDSU researchers have some West River partners with large ranches, most of the farms in the study will be within a one- to two-hour drive from Brookings. About a dozen undergraduate and graduate students will be involved with on-farm sampling, interviews and simulation exercises each year.
The goal is not only to use the precision ag technologies to reduce nutrient leaching, but also increase crop and livestock productivity, Gardezi noted. In Vermont, the sensor technologies will seek to reduce the environmental impact of phosphorus, while the focus will be on nitrates in South Dakota.
What the researchers learn can lead to tracking how much carbon is being stored in soil, which, for instance, may help South Dakota corn ethanol meet California’s low carbon fuel standard.
Offering incentives, evaluating models
Participating farmers will collect data on nitrates and/or phosphorus using aerial drone sensors and ground-based sensors deployed across their fields. Each participant will receive a basic monetary incentive to enroll in the study.
The farmers will then be divided into four groups. The control group will analyze its data using an established agronomic model. “This will be the baseline group,” Gardezi noted. Another group will use the established model and then earn additional dollars based on the extent to which it reduces nitrate leaching.
The other two groups will receive a new computational model developed through deep learning and AI-based algorithms that also takes into account data privacy and will also receive training on how to use the new model. In addition, one of these groups will receive performance-based money for reducing nitrate leaching.
“We want to figure out whether farmers’ trust in new precision ag tools is developed through a better, more secure computational model, enhanced hands-on training, performance-based incentives or via implementing all of the above,” Gardezi said. The study results will help researchers develop workshops to train farmers and curriculum for precision agriculture students.
“Eventually, we hope that this level of care and responsibility in the innovation process can build stronger human-machine networks that will result in greater sustainability for agriculture nationwide,” he concluded.