What crop advisors really want from AI tools: evidence from a North American study
A new study from Virginia Tech and the University of Vermont, conducted with the American Society of Agronomy, delivers one of the first large-scale empirical tests of how Certified Crop Advisors (CCAs) judge AI-enabled decision support systems (AI-DSS). Published in Technological Forecasting and Social Change, it pinpoints which design choices drive adoption-and which create friction.
Using a discrete-choice experiment, the team quantified how advisors trade off cost, accuracy, spatial precision, and data ownership. The result is a clear hierarchy of needs that developers, researchers, and policy teams can act on now.
What drives adoption
- Simplicity beats sophistication: Advisors preferred easy-to-use tools-especially those that tap satellite data-over ultra-accurate systems that demand heavy data inputs.
- Trust hinges on cost and data control: Transparent pricing and user-retained or shared data ownership were decisive. Black-box data policies reduce interest, even when performance is strong.
- Augment, don't automate: The most valued features were editable recommendations, local calibration, and options for field verification. Advisors want AI to support professional judgment, not replace it.
- Attitudes matter: Advisors optimistic about AI were more open to data-intensive tools; advisors with privacy concerns were wary of systems requiring extensive farmer data.
As study lead Maaz Gardezi put it: "Technical performance of AI tools matters in agriculture, but cost and data ownership-especially shared or open models-are pivotal to selection. Crop advisors prefer systems that augment rather than replace professional judgment."
Why this matters now
AI already influences fertilizer timing, pest and disease management, irrigation, and carbon/nutrient accounting. Yet adoption lags on many mid-sized and smaller farms due to concerns about affordability, privacy, and transparency.
"Certified crop advisors are among the most trusted technical experts that farmers in the US turn to," said Asim Zia of UVM. "Designing AI decision tools that enhance, not replace, their expertise is essential for building agricultural systems that are productive, equitable, and climate-resilient."
A socio-technical playbook for trustworthy ag AI
The authors argue for a socio-technical approach-aligning algorithms with the practical constraints and values of the users who carry the risk. Their recommendations:
- Co-create with CCAs and farmers from the outset (requirements, prototyping, validation).
- Make cost structures explicit and communicate trade-offs clearly (accuracy vs. inputs, precision vs. usability).
- Adopt user-controlled data governance (retain, share, or open models with clear terms and portability).
- Keep humans in the loop-editable outputs, local calibration, and workflows that respect advisor autonomy.
"These insights help move AI for agriculture beyond performance metrics," noted co-author Donna Rizzo. "The goal is trustworthy, context-sensitive tools that work for diverse farms and advisory systems."
Practical takeaways for science and research teams
- Researchers: Integrate discrete-choice experiments early to quantify feature trade-offs. Measure UI friction, calibration burden, and time-on-task alongside accuracy.
- Developers: Default to minimal required inputs and clear satellite-data integration. Ship explainability, editability, and offline/field verification modes.
- Advisors and ag retailers: Pilot with small grower groups, document privacy safeguards, and negotiate shared/open data models where possible.
- Policy and funders: Tie incentives to co-creation, data control, and transparency. Support extension-led validation networks for independent testing.
Study details and further reading
- Peer-reviewed paper: Technological Forecasting and Social Change (DOI: 10.1016/j.techfore.2026.124601)
- About the CCA program: American Society of Agronomy - Certified Crop Adviser
Related resources
Your membership also unlocks:
