By: Guest Writer, Dr. Lynn Sosnoskie, Assistant Professor of Weed Ecology and Management in Specialty Crops at Cornell Agritech
Anyone who has heard me speak knows that I see great promise in the integration of artificial intelligence (AI) with advanced weed control technologies like laser weeding and vision-guided spraying. These tools offer the precision needed to reduce herbicide use while maintaining effective weed suppression. AI-driven platforms can distinguish between crops and weeds in real time, enabling site-specific management that minimizes crop injury potential and environmental impact. I’m particularly excited about using these technologies to address herbicide-resistant weeds and other difficult-to-control species through novel means.
But AI has its drawbacks. The information it provides is not always accurate. For example, my initial experiences with the Carbon Robotics Laser Weeder and the Verdant Robotics Sharp Shooter in New York were marked by poor control of common ragweed, a species not commonly found in the western US. and therefore not represented in the machines’ training algorithms. In another instance, I asked ChatGPT to generate a description of Palmer amaranth, and the output incorrectly stated that the species has wind-dispersed seed. Palmer amaranth seed lacks a pappus, the umbrella- or parachute-like structure found on dandelions that enable long-distance transport on wind currents. This highlights the importance of reviewing AI-generated content for accuracy.
A recent article by Dr. Eric Prostko (University of Georgia) in Farm Progress highlights the risks of relying on AI recommendations for pest management, using real-world examples where such advice could have caused problems for the producers. https://www.farmprogress.com/crop-protection/extension-specialist-beware-of-ai-herbicide-recommendations Specifically, AI tools suggested the use of herbicides that are not legally registered for use on the target crops, leading to regulatory violations as well as the potential for severe crop injury (for example, an atrazine recommendation in peanuts).
While AI tools may seem convenient, they lack the local agronomic knowledge and judgment that trained professionals bring to pest management decisions. Cornell Cooperative Extension has an extensive network of specialists deployed across the state to assist growers with their weed control needs. They are here to help you, so please take advantage of their experience when it comes to developing pest management plans. And, as always, CONSULT THE LABEL before making any pesticide applications. The label provides critical information for safe and effective use. Applying a pesticide in a manner inconsistent with its labeling can result in crop injury, environmental harm, and potential health risks to applicators or bystanders.