In recent years, artificial intelligence (AI) has steadily found its way into diverse sectors of society, transforming the way decisions are made across industries. Agriculture, as a cornerstone of human civilization, is no exception. A groundbreaking study co-authored by researchers from Virginia Tech and the University of Vermont provides one of the first comprehensive, large-scale empirical examinations of how Certified Crop Advisors (CCAs) across North America perceive and evaluate AI-enabled decision support systems (AI-DSS) tailored for agriculture. Published in the prestigious journal Technological Forecasting and Social Change, this research sheds new light on the nuanced factors influencing AI adoption in agricultural advisory services.
The interdisciplinary research team, which included experts from public policy, environmental engineering, computer science, and agricultural sciences, collaborated closely with the American Society of Agronomy. Utilizing a discrete-choice experiment—a sophisticated method for understanding decision-making by evaluating trade-offs—the study dissected the preferences of CCAs regarding AI-DSS features such as cost, accuracy, spatial precision, and data ownership. The results reveal a complex interplay between technical performance and trust-related factors, fundamentally challenging simplistic assumptions that accuracy alone drives AI adoption in agriculture.
Perhaps the most remarkable insight emerging from this study is that simplicity and usability trump ultra-high accuracy when it comes to AI tools designed for crop advisors. CCAs showed a strong preference for systems that were intuitive and easy to use, particularly those integrating readily available satellite data. Conversely, AI-DSS offerings requiring intensive data inputs and delivering marginal increases in accuracy were less favorably viewed. This finding underscores the importance of building AI solutions that fit smoothly into the demanding and varied workflows of agricultural professionals rather than imposing cumbersome technological burdens.
Another pivotal discovery concerns trust, intricately linked to transparency and governance of data. Far beyond the financial cost of AI tools, issues around data ownership emerged as critical determinants of whether crop advisors would adopt these technologies. The study highlights a pronounced preference for systems that allow users to maintain full or shared control over their data, reflecting deeply rooted concerns about privacy and ethical governance. This preference for open or shared data models signals an urgent need for AI developers to establish transparent, user-centered data policies that respect the autonomy of both advisors and farmers.
Crucially, the study reveals that crop advisors are not interested in relinquishing their professional judgment to machines. Rather, they see AI-DSS as complementary aides that should augment their expertise without fully automating complex decisions. Advisors valued tools providing editable recommendations, the ability to calibrate systems locally, and options for field verification—elements that preserve human-in-the-loop integrity. Such design features bolster the ability of crop advisors to contextualize AI insights and adapt them to heterogeneous field conditions, seasons, and socio-economic realities, thereby anchoring technology within practical, on-the-ground expertise.
The researchers also found that individual attitudes towards AI play a significant role in adoption likelihood. Advisors who maintain an optimistic outlook toward AI were more willing to engage with data-intensive systems, suggesting that perceived benefits can offset concerns about complexity or data privacy. Conversely, those with entrenched privacy worries tended to shy away from tools demanding extensive farmer data. This attitudinal divide points to the necessity of building awareness and trust, alongside technical improvements, to foster wider acceptance of AI technologies in the agricultural advisory domain.
Lead investigator Maaz Gardezi of Virginia Tech encapsulates the implications succinctly: while technical performance is undeniably important, factors like cost and data ownership—especially models fostering shared or open access—are pivotal to technology selection by CCAs. This nuanced perspective defies simplistic metrics that prioritize mere accuracy, emphasizing instead the socio-technical context in which agricultural AI tools operate. The findings advocate for a paradigm shift that sees AI as an enabler of expertise, not a substitute, harmonizing machine intelligence with human insight.
This research arrives at a time when AI-powered models increasingly influence critical farm management decisions, ranging from precise fertilizer application to pest and disease control, irrigation scheduling, and even carbon and nutrient accounting. Despite these promising capabilities, widespread adoption among mid-sized and smaller farms remains elusive. The study sheds light on the underlying causes—issues of affordability, privacy, transparency, and trust—that complicate the path from innovation to practice. By focusing on the gatekeepers of agricultural knowledge, certified crop advisors, the research presents a realistic lens through which to understand adoption dynamics.
University of Vermont professor Asim Zia emphasizes the significance of this approach: “Certified crop advisors are among the most trusted technical experts that farmers in the US turn to.” Designing AI tools that enhance rather than supplant their expertise is fundamental for fostering agricultural systems that are not only productive but also equitable and resilient in the face of climate challenges. This highlights a socio-technical imperative for AI development: grounding algorithms within the lived realities and values of human users rather than abstract computational ideals.
To that end, the authors propose a socio-technical framework for trustworthy AI in agriculture. This framework advocates co-creation with end users—in this case, crop advisors and farmers—from the earliest stages of development. It also calls for clear and transparent cost structures, communicable trade-offs regarding accuracy and data use, and governance models that prioritize user control over data. Crucially, it advances “human-in-the-loop” designs that preserve advisor autonomy, preserving the delicate balance between technological assistance and professional discretion. This framework, the authors argue, paves the way for AI tools that are not merely performant but also context-sensitive and trustworthy.
Professor Donna Rizzo of UVM, co-author of the study, articulates the broader vision: moving AI for agriculture “beyond performance metrics” to create tools that genuinely work for diverse kinds of farms and advisory systems. Such tools must be adaptable, respectful of privacy, and attuned to the patchwork of ecological and social conditions that characterize modern agriculture. This vision challenges AI developers, policymakers, and funders to rethink the metrics and incentives that drive innovation, prioritizing usability, transparency, and equity alongside raw computational power.
The study, entitled “A socio-technical framework for analyzing crop advisors’ preferences for AI-based decision support systems,” will appear in the May 2026 issue of Technological Forecasting and Social Change. Funded by the National Science Foundation and the USDA National Institute of Food and Agriculture, the research exemplifies a collaborative effort across institutions and disciplines to anchor AI development in the needs and values of agricultural practitioners. Ultimately, this work charts a forward-looking course for AI integration that respects human expertise, safeguards data sovereignty, and fosters resilient, sustainable food systems.
By spotlighting the voices of the people who advise farmers daily, this research offers a fresh vantage point from which to rethink AI’s potential in agriculture. It challenges technologists and stakeholders to transcend narrow conceptions of accuracy and efficiency, embracing a more holistic vision that balances innovation with trust, simplicity with functionality, and technological progress with ethical stewardship. In doing so, it illuminates pathways toward agricultural AI systems that empower rather than overshadow those who cultivate the land and feed the world.
Subject of Research: People
Article Title: A socio-technical framework for analyzing crop advisors’ preferences for AI-based decision support systems
News Publication Date: 24-Feb-2026
Web References: https://doi.org/10.1016/j.techfore.2026.124601
References: National Science Foundation (Grant Nos. 2202706 and 2026431); USDA National Institute of Food and Agriculture (Award No. 2023‑67023‑40216)
Image Credits: UVM College of Agriculture and Life Sciences
Tags: agricultural AI tools usabilityAI adoption in agricultureAI decision support systems agricultureAI trust factors in farmingAI-DSS cost and accuracy trade-offsAI-enabled crop advisory servicesCertified Crop Advisors perceptionsdata ownership in AI agriculturediscrete-choice experiments agricultureinterdisciplinary agricultural researchspatial precision in agricultural AIVirginia Tech agricultural AI study
