Using ChatGPT on the Farm

OMAHA (DTN) -- As artificial intelligence tools such as ChatGPT continue expanding their capabilities, at least a few farmers are finding ways to integrate artificial intelligence (AI) products into daily work to help make quick calculations or query what they see on the farm.

Arkansas farmer Travis Senter II explained a little bit about how he uses ChatGPT when he spoke last month to participants in the Conservation Technology Information Center (CTIC) tour. Having signed up for the AI tool when it came out in late 2022, Senter said it's become a reliable tool for finding quick answers.

"I use it almost every day in some fashion," Senter told DTN in an interview. "There are lots of ways you can use this information every day that's available at your fingertips. If I can ask it and it spits it out in ten seconds, that saves me time."

Rather than using the free version, Senter pays $20 a month for ChatGPT, which allows him to upload documents, as well as create his own GPT (generative pre-trained transformer) that caters to Senter's interests and farming business.

"It's not going to know everything that's going on in your life, or going on locally or on a global level, but it may know more than what you fathom."

Talking about doing the math of how much fuel was in a tank, Senter said he could have found a website that would have had him plug in multiple calculations. "I didn't want to go through the process of figuring out the equations. I had the basic measurements and I said, 'I've got so many inches of fuel in it, and it just spits the answer out."

Senter added, "Those are real-world things that I don't have time to sit there and calculate all of this. I need an answer quick, and it gives it to you."

Traditionally when a farmer finds new weeds in a field, they need to call or send an image to an agronomist. The AI tools now will identify the weed and suggest spraying options for it.

"I know a lot of weeds, but I'm not a weed scientist," Senter said. "But I can take a picture of a weed in a field and upload it. I used it the other day. It told me exactly what it was, and what I could spray on it. Obviously, you need to read the label and all of that, but it can even download the label for you. There are a lot of things it does for you without having to go anywhere else."

Right now, it's unclear just how many farmers use various AI tools in their operations. Answering a query from DTN, ChatGPT stated precise statistics on the number of U.S. farms using AI tools are not well documented. Overall adoption rates for AI in businesses remain relatively low at roughly 3.8%, ChatGPT stated. "For farmers, integrating AI tools can mean more efficient farm management, better yield predictions, and improved decision-making capabilities, ultimately leading to enhanced productivity and sustainability in their operations," ChatGPT responded.

Even getting emails from brokers and grain merchants that send tips on trading, Senter said he loads those into Chat GPT just to get a summary of the information in the shortest method possible.

"Even farm bill programs, I throw those in and let it sort of summarize things," Senter said. "That way you don't have to sit there and comb through it. You can just ask it questions as you need it. The biggest thing is that I'm not judged on the questions that I ask. I can ask whatever I want, be it a kindergarten-level or a complex question. I'm not judged, and I don't get the answer in a judgmental fashion."

Senter also began experimenting with the program a lot more last summer after he had seen other people creating their own models and uploading files. He started loading more historical information from his own fields such as yields, irrigation data and planting dates. He also began adding local weather models into his GPT. Senter said he will then ask the program to provide pollination dates and daily temperatures for that time.

"I'm trying to predict things like why my yield was better one year versus the other," he said. "On some fields it was actually pretty dang close. Some fields it was off, which may have been differences in irrigation, but it sort of averages everything out and you get close."

Better yield and production forecasts on specific fields also helps farmers consider how much they should forward contract each year, for instance.

"I think that has a big place in the future of agriculture -- just trying to do some real-time predictive work with your own information," Senter said.

He added, "It's only as good as the data that you feed it, so if you feed it false information then it's going to give you false answers."

These programs typically try to average everything out. Senter said it can sometimes take multiple versions of questions to refine some answers, especially when you have several variables that add complexity to the predictions.

"You can argue with it, saying that something just doesn't sound right. 'Tell me why you decided to use this?' And you just keep following that strategy with these text models and let it sort of guide you to how it got the answer.

"It's always about the prompts and how you ask questions. That's something I've learned I had to get better at. The prompts you give it are sort of what feeds the answer. On Google, you get one question and that's it. With this, you can refine what you are asking to get the answer you need to have."

Senter said there is a "double-edged sword" of data privacy concerns, especially if someone is pulling data stored in a cloud to draw an answer from ChatGPT.

"I don't feed it anything I wouldn't mind sharing with anyone."

ChatGPT, when queried, noted, "Users must be cautious about the information they share with any cloud-based service."

But when the data is there, it may just take a question or two to pull up specifics from that field last year rather than going through multiple folders and files on a desktop. As his family and workers on the farm operation become more adept at using ChatGPT, Senter said they don't have to specifically be tech savvy to query the AI platform.

"My father doesn't know how to turn on a computer, but he knows how to work his phone," Senter said. "He knows a lot of things and he also knows what questions he wants answers for, and he relies on me for answers to that. But if he can pick up his app and find out in one second what his yield was on a certain field last year, that changes how information is delivered to people."

Farmers are likely to see more specific applications develop as well as tools to help them use AI, especially if land-grant universities have a say. Purdue University operates the Open Ag Technology and Systems Center (OATS). The University of Illinois, for instance, has the Center for Digital Agriculture that has been in business since 2018. Kansas State University last September created the Institute for Digital Agriculture and Advanced Analytics (ID3A). The University of Missouri earlier this year also launched the Digital Agriculture Research and Extension Center (DAREC).

Looking at areas ranging from irrigation to marketing, Senter said he expects AI tools will start opening up a lot more avenues for farmers to get answers they need on the farm without seeming judged for their questions.

"I think it's got a huge place and potential for what we can do with our data and what we collect from the agricultural side," Senter said. "It's going to change the way you approach things."

Also see, "Artificial Intelligence Tools Like ChatGPT Could Change Decision Making on the Farm," https://www.dtnpf.com/…

Chris Clayton can be reached at Chris.Clayton@dtn.com

Follow him on social platform X @ChrisClaytonDTN

(c) Copyright 2024 DTN, LLC. All rights reserved.