Can generative AI make farming decisions? Current status and future pathways – A case study in row crop production with ChatGPT

Abstract

The agricultural decision-making process is experience-based, knowledge-dependent, time-sensitive, complex, and driven by historical data. Planting, fertilization, irrigation, and chemigation are key categories in farm decision-making, and currently, there is no one-shot decision-support tool that covers all these activities. Generative Artificial Intelligence (AI) models are more advanced than traditional machine learning and deep learning models. These models have been trained on vast amounts of data from the internet, allowing them to accept unstructured data in various forms and generate human-like text, solutions to problems, and scenario predictions. Given this capability, we became interested in exploring the potential of generative AI in agricultural decision-making. We designed a study to evaluate how well these models can make management decisions in a row crop production environment with humans in the loop. The study began in March 2024 on sprinkler corn plots in North Platte, NE, managed by the TAPS (Testing AG Performance Solutions) program at the University of Nebraska-Lincoln. We evaluated the ChatGPT-4o generative AI model, developed by OpenAI, in terms of its ability to generate decisions for seed selection, cover crop termination, fertigation, irrigation, and chemigation in real time. The model's input included unstructured past management decisions from the TAPS program, 2024 pre-plant soil health lab reports, farm management decision request emails from the TAPS program manager throughout the growing season, and the latest sensor-based weather and soil water content data relevant to each decision category. We found that the decisions made by ChatGPT-4o were faster than human decision-making and were logical, practical, and executable. Notably, the plots managed by AI ranked number 8 in yield and 13th by agronomic efficiency among all 31 plots managed by experienced growers. These findings are promising and suggest that generative AI models can build generalized farm decision-support tools for row crop production. • The potential of ChatGPT-4o for agricultural decision making in corn production was explored. • Decisions included cover crop determination, planting, fertilization, irrigation, and insecticide application. • ChatGPT-4o accepted multi-modal inputs including historical data, NDVI maps, and sensor data. • Decisions made by ChatGPT-4o were logical, practical, executable, and faster than human decision-making. • Generative AI models have great potential to build generalized farm decision support tools for row crops.

Ähnliche Arbeiten