Assessing the Impact of AI on Agriculture in Remote and Resource Poor Areas

Dr. Ataharul Chowdhury , associate professor in the University of Guelph’s School of Environmental Design and Rural Development within the Ontario Agricultural College (OAC), was born and raised on a family farm in Bangladesh. His early experiences working with the local farming community led to his research today at the intersection of artificial intelligence and agriculture.?

Having worked across the globe, Chowdhury aims to understand how we can best leverage AI-based technologies in the pursuit of sustainable precision agriculture in remote and resource poor areas and the ethical and legal challenges these technologies present. Chowdhury’s approach is holistic, focusing on both farmer capabilities with these technologies, as well as the policy and good governance practices that must be in place to ensure effective and responsible use of them and to combat misinformation.

Chowdhury’s research makes his affiliation with U of G’s Centre for Advancing Responsible and Ethical AI ( CARE-AI ) a natural one. Not unique to the agricultural sector, the concerns of AI development and use will be familiar to many – data ownership, privacy, wrongful recommendations, misinformation, and algorithmic bias.

In 2020, Chowdhury was awarded a CARE-AI seed fund grant to pursue these concerns. With collaborators Mahatab Uddin and Ashad Kabir, Chowdhury’s funded research identifies areas of ethical and legal concern for deploying AI in climate-smart agriculture and make recommendations for how associated harms should be addressed.?

One of the areas of concern is data inaccuracy. Sensors feed real-time data analyzing soil moisture, salinity, pH, and nutrient levels for classification and outlier detection, and drones facilitate the computer vision monitoring of crop counts and weed detection. This data is used by variable rate technologies (VRTs) to tailor the dispersion of water, fertilizer, and chemicals across a field. If these technologies are deployed with poor, limited, or biased data sets (more likely the case on smallholder farms) resulting recommendations may damage soils and reduce crop yields, resulting in financial hardship for farmers and generating distrust of these systems.?

Collaboration with farmers is essential to preventing harmful outcomes. Data sets should be integrated and validated with local knowledge. From the perspective of Chowdhury’s holistic approach, a necessary condition for the success of AI to innovate sustainable agriculture are the appropriate laws to deal with potential harms. This is discussed in the team’s paper entitled “Legal and ethical aspects of deploying artificial intelligence in climate-smart agriculture”, published in AI and Society.

Outside of research, Chowdhury enjoys gardening, playing sports - especially cricket and organizing social gatherings and community initiatives. His overall philosophy: as community members, on both local and global scales, we owe it to each other to share knowledge, experience, and a helping hand in the pursuit of a better world.