AI Decision Making in Agriculture
Like the invention of electricity or the internal combustion engine, the development of artificial intelligence (AI) is expected to radically change the economy and society. Production agriculture will not be immune. Already AI in drones or satellites monitor fields detecting nutrition stress and pest outbreaks and suggest prescriptions. AI sprayers spot weeds and direct spray at the correct rate. Weeding robots and tractors use AI for guidance. AI’s use in agriculture will only grow.
One area where AI could grow is as a producer decision support tool. Cory Walters, an agricultural economist with the University of Nebraska, recently wrote about a study conducted as part of the University’s Testing Ag Performance Solutions (TAPS) farm management competition on the usefulness of generative AI as a decision-support tool in crop production. TAPS is a farm management competition where participants hone their crop management and marketing skills, vying for prizes emphasizing profitability and efficiency. Producers, industry leaders, students, scientists, and government regulators make up the teams. Last year an AI-assisted decision-making system using ChatGPT guided the decisions of one of the teams. The system was provided with “farm-level data, including soil conditions, weather information, and historical performance” and then generated production decisions. Grain marketing decisions were not made by AI.
Walters wrote the “AI-generated recommendations were logical, timely, and operationally feasible” and the AI-managed plot “achieved above-average yields, ranking in the top third of all participants and producing statistically higher yields than the average farmer-managed plots.” However, the AI-managed plot did not score as well on input efficiency or profitability. For example, irrigation water-use efficiency was lower than average.
Walters said the findings suggest AI can effectively support producer production decisions, but it still has limitations. Integrating AI production decision systems with economic optimization tools and market information systems could help strengthen AI’s input efficiency and profitability capabilities. Moreover, because AI relies on the availability of quality data, continued investments in data infrastructure, sensor networks, and broadband connectivity are needed. Also, the study showed the importance of human oversight to AI decision-making. Thus, AI is more likely to augment producer decision-making rather than replace it.
A recent survey of 166 farmers by MorganMyers, an agricultural marketing firm, and Ag Access, an agricultural market researcher, found 48% of farmers surveyed said they use AI tools on a weekly or more frequent basis to assist in decision making. But 45% said they were not comfortable using AI and 55% of crop farmers surveyed reported low or no use of AI. Dairies, large-scale operations, and farmers under the age of 35 had higher usage rates. All signs point to the growing adoption of AI in agriculture. AI promises greater efficiency, improved management tools, and potential cost savings. But it also comes with the costs of adopting new technologies and new ways of approaching management. Not an easy decision for farmers in today’s challenging financial environment. For further information on the study, go to: https://agecon.unl.edu/can-ai-improve-farm-decision-making-evidence-and-tradeoffs/