AI Technology in Small Farm Agricultural Decision Support Systems

The Agriculture Improvement Act of 2018 (2018 Farm Bill) includes provisions that address the unique circumstances and concerns of socially disadvantaged, beginning, limited resource, and veteran farmers and ranchers (collectively referred to as "historically underserved producers"). Since the outbreak of COVID-19 in 2020, there has been a significant increase in interest among socially disadvantaged and underserved audiences (both young and old) in farming to provide supplemental income and address food insecurity. Many of these farmer groups lack adequate farming experience, have little start-up capital, and have limited access to financial credit and land.

Nazia ArbabThe University of Maryland Eastern Shore (UMES) Extension conducted a Farmers’ Needs Assessment in Agriculture (FNAA) in 2021/2022. The primary goal of the FNAA was to optimize the impact of Extension programs on the socio-economic and environmental benefits for the target audience. Based on the data collected, farmers reported various problems associated with farming, which were grouped into three tiers based on the percentage of responses. The top challenges identified in Tier 1 included a lack of capital (93%), expensive production inputs, and a labor shortage (92%). In Tier 2, recurring themes that surfaced at 90% were a lack of produce-processing facilities, high interest in credit, lack of access to market outlets, and a lack of computer knowledge and skills. Tier 3 identified a lack of farm business planning skills (88%) as one of their chief concerns.

Due to its high cost, AI is usually deployed on large monoculture or high-value crops, even though most agriculture occurs on small and midsize farms (USDA, 2017). According to the 2017 Census of Agriculture, an increasing number of female, first-generation, and American Indian farmers are key players in modern agriculture. The innovations developed by AI need prioritization to enable AI methods for the 98% of US farms that are family-owned (i.e., operated by individuals related to the owner) and, in particular, the 94% of all farms that are classified as small and midsize, which account for over 40% of the value of total agricultural production.

Therefore, this study attempts to answer the following research question: To what extent can artificial intelligence techniques help increase the farm livelihood for socially disadvantaged and underserved farmers? To respond to this question, this paper proposes a comparative approach to how predictive and generative AI technology in decision support systems can generate economic outputs.