Contributed by Albert Adjesiwor, University of Idaho
To say artificial intelligence (AI) has taken us by storm is an understatement. AI has become integral to our everyday lives, and it is safe to say it is here to stay. This technology will revolutionize how we work, our interaction with technology, among other things I can’t even begin to think about. I would have to admit, I am no expert on AI. However, like most of us, I have been wondering how AI would affect my work going forward. More specifically, how would AI affect my interaction with extension clientele? To answer this question, I have been assessing how prominent AI chatbots would respond to a few common questions farmers have or may ask.
There are several AI chatbots available but for this article, I will focus on the “big three” and Norm, an AI developed specifically to provide answers to common agricultural questions. I left out Apple Intelligence because I rarely use Apple products and didn’t have enough time for a deep dive before writing this article.
Figure 1. The “big three” chatbots (ChatGPT, Copilot, and Gemini) and Norm, an AI developed to provide answers to common agricultural questions.
A little about these AI chatbots
Copilot
Copilot is a chatbot developed by Microsoft that you can interact with. Copilot allows you to choose from three styles of conversation: more creative, more balanced, or more precise. The style you select will determine how detailed the copilot answers any given question.
Gemini
Gemini, formerly known as Bard, is a generative AI chatbot developed by Google and launched in 2023. Gemini understands text, images, videos, and audio and is described as capable of completing complex tasks in math, physics, and other areas.
Norm
Norm is Farmers Business Network®’s AI Ag Advisor built based on ChatGPT’s language model. Norm is being trained on high-quality agronomy content with the aim of helping farmers and agronomists answer questions about agronomy, farm management, and beyond. Norm is still being Beta tested, meaning it is being released to a small group of users to evaluate its performance in a real-world setting before it is made widely available.
ChatGPT
Well, this one probably needs no introduction. The previous three chatbots mentioned have been built from the ChatGPT model. OpenAI developed ChatGPT as a chatbot that uses AI to generate human-like responses to user prompts. It uses machine learning to process data and understand natural language. As such, it can respond to questions and write content like essays, articles, emails, social media posts, among others.
To test the capability of AI chatbots to answer common agricultural questions, I asked each chatbot three questions. I categorized question 1 as a “statement of fact”, and questions 2 &3 as “recommendation questions”. I didn’t ask for a region-specific answer so these answers are not meant to be recommendations for the Pacific Northwest. The answers below are almost verbatim from what the chatbot provided.
Question 1. How long after zidua herbicide application can I safely plant grasses grown for seed?
| Chatbot | Answer | Accuracy |
| ChatGPT | After applying Zidua herbicide, it’s recommended to wait four months before planting grasses grown for seed…the four-month interval ensures that any residual herbicide in the soil degrades sufficiently to avoid damage to the germination and growth of the grass seed.
|
❌ Incorrect |
| Gemini (Bard) | Generally, it’s recommended to wait at least 30 days after applying Zidua before planting grasses grown for seed. However, this is just a general guideline
|
❌ Incorrect |
| Copilot | After applying Zidua herbicide, it’s generally recommended to wait 2 to 4 weeks before planting grass seeds. This allows the herbicide to fully dissipate and ensures the soil is safe for new growth.
|
❌ Incorrect |
| Norm | The rotational crop interval for planting grasses grown for seed after applying Zidua® SC herbicide is dependent on the amount of precipitation received. If more than 15 inches of precipitation (rainfall/irrigation) has occurred from the time of application to planting, the interval is 11 months. If less than 15 inches of precipitation has occurred, the interval extends to 18 months.
|
✅ Correct |
Question 2: Herbicide recommendation to control Italian ryegrass in spring wheat
| Chatbot | Answer | Accuracy |
| ChatGPT | Here are some effective herbicides commonly recommended for controlling Italian ryegrass in spring wheat:
Axial XL (Pinoxaden), Osprey (Mesosulfuron-methyl), PowerFlex (Pyroxsulam); Axiom DF (Flufenacet + Metribuzin), Zidua (Pyroxasulfone)
|
🔶 Mostly correct |
| Gemini (Bard) | To control Italian ryegrass in spring wheat, consider using an herbicide that specifically targets this weed without harming the wheat crop. Here are some options to explore:
Trifluralin, Pendimethalin, Dicamba, 2,4-D.
|
❌ Mostly incorrect |
| Copilot | To control Italian ryegrass in spring wheat, it’s recommended to use a residual herbicide in a post-emergence spray, especially if you haven’t used a pre-emergence herbicide at planting. One effective option is pyroxasulfone, which can help reduce Italian ryegrass emergence when applied in a delayed pre-emergence application. Timing is critical for the best results.
|
✅ Correct |
| Norm | Zidua® SC herbicide: This herbicide contains the active ingredient Pyroxasulfone and is labeled for use with spring wheat to control Italian ryegrass.
RedEagle Flumioxazin 51% WDG Specialty + Field (California): This herbicide contains Flumioxazin and is also labeled for use with spring wheat to control Italian ryegrass.
|
🔶 Mostly correct |
Question 3: Herbicide recommendations to control ACCase resistant wild oat in spring wheat
| Chatbot | Answer | Accuracy |
| ChatGPT | Below are recommended herbicides and strategies for controlling ACCase-resistant wild oat:
Osprey (Mesosulfuron-methyl), PowerFlex HL (Pyroxsulam), Axiom DF (Flufenacet + Metribuzin), Zidua (Pyroxasulfone), Rimfire Max (Mesosulfuron-methyl + Propoxycarbazone-sodium) Group 3 Herbicides (Triallate)
|
❌ Mostly incorrect |
| Gemini (Bard) | Trifluralin, Pendimethalin, Aryloxyphenoxypropionic acid (AOA) herbicides: These herbicides, like quinclorac or clodinafop-propargyl, can be effective against ACCase-resistant wild oat, but resistance to these herbicides is also emerging. Pyroxsulam: This herbicide can provide control of ACCase-resistant wild oat, but it’s important to follow label instructions carefully and consider tank mixing with other herbicides for better control.
|
❌ Mostly incorrect |
| Copilot | To control ACCase-resistant wild oat in spring wheat, you can consider using post-emergence herbicides that are effective against this weed. Some options include:
ACCase inhibitors: Clodinafop, Fenoxaprop, and Pinoxaden ALS inhibitors: Everest 2.0, GoldSky, and Rimfire Max3 It’s important to apply these herbicides when all the wild oat seeds have emerged for the best results
|
❌ Mostly incorrect |
| Norm | To control ACCase-resistant wild oat in spring wheat, you have a couple of herbicide options available from FBN: Zidua® SC herbicide: This herbicide contains the active ingredient Pyroxasulfone and is labeled for use with spring wheat to control wild oat.
Pendimethalin 3.80 Value Pick: This herbicide contains Pendimethalin and, based on similarity with labeled products, may be labeled for use with spring wheat to control wild oat.
|
❌ Mostly incorrect |
To be honest, I was expecting these chatbots to be all over the place in their answers to these questions. Although the majority of the responses were incorrect or not close to what an experienced extension specialist or agronomist would say or recommend, we can’t downplay how good these new AI chatbots are at understanding and responding to questions like these. It was not surprising that nearly all the AI chatbots got question 3 wrong. Herbicide mode of action and resistance management are tricky even for professionals. In crops such as wheat with very limited herbicide options for grassy weed control, resistance to group 1 (ACCase) herbicides becomes even harder. It is worth mentioning that most of these AI chatbots were launched barely 2 years ago. In human terms, these are supposed to be like “second graders”. I expect these to get extremely good at answering agricultural questions as they are trained with more data. These AI have the added benefit of having access to any publicly available historical and future data or publications so yes, we should expect them to be very good in the near future. For now, if you are or plan on using some of these chatbots for answering agricultural questions or herbicide recommendations it is advisable to confirm any answers you get from chatbots with a specialist. To conclude, there’s no doubt AI will play a huge role in agricultural extension in the future. In-person events and personal interactions would be more crucial in agricultural extension going forward to maintain the relevance of agricultural extension.