Artificial Intelligence Could Create New Opportunities for Sustainable Agriculture, with Dr. Nathan Boyd

Artificial intelligence has been a hot topic in the business world over the last year.

Some of it has been really interesting, and some has been just hype.

But in the field of sustainable agriculture, where we already work with massive volumes of data, there are some A.I. applications that could provide real value to growers.

So in this episode, we meet Dr. Nathan Boyd, the Associate Director of the University of Florida Gulf Coast Research and Education Center.

He's helping develop Florida's state-funded Center for Applied Artificial Intelligence, and with a background in agriculture and weed control, knows that new technology has to create a clear value for growers if it's going to succeed.

We'll discuss how A.I. is different from other new technologies in agriculture, how it could be used to create value, and why it could be a game-changer for sustainability.

Episode Transcript

Nathan Boyd

The strength of artificial intelligence is the capability to take large volumes of data and extract from that, actionable outputs, which is things that you could actually help the grower know what needs to be done.

Dusty Weis

Welcome to The FARMSMART Podcast, presented by Nutrien Ag Solutions, where every month we're talking to sustainable agriculture experts from throughout the industry. As the leading source of insight for growers and evolving their sustainability practices while staying grounded in agronomic proof, FARMSMART is where sustainability meets opportunity.

Sally Flis

We don't just talk change. We're out in the field helping you identify the products, practices and technologies that bring the future to your fields faster. I'm Dr. Sally Flis, Director of Program Design and Outcome Management.

Dusty Weis

And I'm Dusty Weis, and we're joined today by Dr. Nathan Boyd, Associate Director of the University of Florida Gulf Coast Research and Education Center, where he is helping develop a state-funded center for applied artificial intelligence. So, Doctor Boyd, thank you for joining us.

Nathan Boyd

My pleasure. Glad to be here.

Dusty Weis

Really been looking forward to this conversation for a while here because it's super topical. It's been top of mind for a lot of folks. We wanted to talk to you because artificial intelligence has been a hot topic lately. We've been hearing all about the opportunities it's going to open up in different fields, the data it could interpret, the efficiencies it's going to create.

And when it comes to sustainable agriculture, a lot of the time when you're using data to do something more efficiently, you're also doing it more sustainably. So for starters, let's level set a little bit. What is this AI that you're exploring in this project, and how is it different from the sort of agricultural computerization and automation that we've seen in the past?

Nathan Boyd

That's a great question. So AI is just a tool. It allows us to do things more effectively, more efficiently than we could in the past. And just to give some examples, precision agriculture has been around for a really long time. But if you go back and read the literature from a decade or more ago, what you'll find repeatedly is you can see the authors kind of saying, you know, “We wish we could do this.”

And what's happened is a lot of what they wish they could do, we can now do. So we can do things with far greater precision, far more accuracy. We can extract information from much more complex data sets than we could in the past.

Sally Flis

Nathan, when we talk about introducing any of these new technologies into agriculture, there's often pushback from the growers in the field, from people like crop consultants who work for us, that people who are proposing this technology don't really understand the problems we're facing and what we're trying to solve. And then there's always that fear that they're getting replaced by something when it comes to technology.

So it's important to call out that you have come from an agricultural background. Can you tell us a little more about your history and how you got to be working in this AI field for agriculture?

Nathan Boyd

Sure. I've been working in agriculture since I was little, actually, but doing research for over 20 years, and I'm a weed scientist by training. So all of my research was focused around the management of weeds in specialty crops, canola, wheat. 

And when I moved to Florida about 12 years ago, this is when I first started kind of beginning to look at AI and it was a very kind of a small entry because we are looking at this field, and I was thinking about, why is it that I'm having to put herbicides over this entire area when I really only need it in the small parts of the field? 

And I reached out to a cooperator by the name of Doctor Arnold Schumann, and him and I began working together on this technology. 

And it was only about a year after that that we started thinking about AI and what we could do with AI, because that enabled us to not only accurately detect where the weeds were within three dimensional space, but it allowed us to identify for the first time what species was there, not just that it was a weed, but what species it was and therefore what you needed to spray. 

And that was kind of the beginning. And it's evolved dramatically from that point, of course. You mentioned earlier a lot of fears about this adoption of new technology, and maybe we can get into that more later if you wish, but our approach has always been: we're going to design technology that addresses the real needs of growers and is easy to use and contains a lot of the characteristics that they want.

And to do that, we brought growers and farmers into the discussion on “What is it we're going to work on, what does it look like when it's done, what do you need it to look like?” And when you involve people from the beginning, you don't have the same concerns when it comes to adoption.

Dusty Weis

If I can paraphrase a little bit, it sounds like you said about defining what the problems were and then seeking a solution rather than bringing in artificial intelligence as a solution that you just tried to fit into the problems that people were facing. 

Nathan Boyd

That's exactly right. We have tried to kind of flip this and kind of take a different approach. There's been a lot of attempts to bring solutions or technologies into agriculture and sometimes solutions to which there was no problem. So what we've tried to do is flip it and say, okay, we want you to define for us what is the problem?

What is it that you actually need? And then let's figure out what is the best way to solve that problem.

Dusty Weis

Well, it's really cool to hear and certainly exciting to explore the use cases. And that's what we want to do here in a second. But I do want to call out also what a big deal it is to see the state of Florida investing in being on the cutting edge of something like this. 

How did the project of the Center for Applied Artificial Intelligence come about, and what is the state of Florida hoping to accomplish there?

Nathan Boyd

Yeah, this is a great story. So I was actually on a trip to Australia where I went to a Center for Applied Robotics, and I was touring that facility and it just popped into my head, “Why couldn't we come up with something like this for agriculture in Florida?” And I made this little video. It's only, I don't know, ten minutes long.

And I gave it to my boss and it kind of started circulating. And then there was like, “Well, you know, go ahead and explore it and maybe it'll happen.” And we started talking to growers. And I spent a long time just traveling all over the state, talking to anybody that would listen. And when we stated addressing the growers that these are real, we could address some real problems that you have. 

Because with specialty crops, everything is done by hand and we don't have a domestic labor force to do it and be able to compete within a global marketplace and being able to find the labor. Those are all big issues and complicated issues. So what we wanted to do was could we come up with solutions that could help the farms in Florida be more competitive? And we felt the Center for Applied AI could do that. 

And we just had overwhelming support from growers all over the state. And that support led to legislators, congressmen, and so on getting on board and supporting this initiative. And the state got behind and provided the funding that we needed. We got some funding for equipment from the federal government and it just kind of took off from there and took on its life of its own.

And we spent quite a bit of time visiting different centers, not necessarily agricultural ones, but robotic ones that we knew were very productive to try to figure out what worked, what didn't work. How can we make sure that this actually meets the needs of the growers in the state? So it's been an exciting development. 

And yes, it's important to note that it's been a lot of investment go into this to make sure it works.

Sally Flis

So, Nathan, you mentioned in your kind of first example of the use of AI in your research, identifying weeds. And so what are some of the other places where you guys are looking at, and what do you see as opportunities for using AI as Dusty mentioned earlier, there seem to be some great links to more efficient use of products and practices on the acre, which are things that support our sustainability metrics that we're looking for.

Nathan Boyd

Yeah, we have developed technologies for a range of different uses, or are developing is probably a better term. So targeted control of pests in general, whether that's weeds, insects or diseases. So that's both the detection, identification, mapping. That's all very critical because it means you can track where the populations are. You could track what's working, what's not working and it allows you to be more efficient in your approach to management. 

Those are the areas that we started with was with the targeted type of spraying, mapping, detection. Other areas that we're working on are automation for manual tasks, and that can be things like runner cutting and strawberries. We’re looking at ways to mechanically harvest, or at least mechanically assist in harvest for specialty crops. 

We're looking at ways to develop apps for different pest detection. A lot of work on yield estimation. Our center has a big emphasis now on using AI to enhance our breeding programs, to more readily detect traits of interest, things of that nature. 

And we're also doing a lot of stuff with digital twins, which is creating a digital twin of a farm so that we can evaluate technology within a digital world before we have to take it to the real world. It allows us to be a lot more efficient in overall development. So there's a quite a few projects already underway.

Dusty Weis

And if I may, Dr. Boyd, a lot of the projects that you outlined there sound similar in scope to some of the projects that people are trying to tackle with artificial intelligence in other fields right now. And maybe the unifying theme in all of those potential use cases is the fact that very often we're talking about things where there's a lot of data that needs to be aggregated and processed en masse, and then insights from that very large amount of data derived and made usable to human users.

Is that where you see the strength in using artificial intelligence to solve some of these agricultural problems?

Nathan Boyd

Yes, I think that's a good summary. What we're doing is very similar to other scenarios, actually. It's really being able to take, like you said, large volumes of data and extract from that actionable outputs or things that you could actually help the grower know what needs to be done. 

This is a good example. If you think about standard scouting practices in specialty crops, where you can hire a scout, you go through and look for things, for example, a standard scout is going to go through and look at less than 1% of the plants in that field and looking for diseases or insects or other types of damage.

With AI, you can take the equipment that's attached to a tractor that's going it through a field anyway, and look at every single plant. So your ability to detect things is much earlier, much quicker, which means you can have a solution much faster. And so that's really exciting because it means you're going to have a huge amount of data to make better choices.

Dusty Weis

It's a force multiplier, really. And again, I think that's what's got so many people excited about the potential that this technology has to impact the world of sustainable agriculture. So I think we've done a really good job of setting the table with some of the opportunities that this technology presents. 

But as with any revolutionary development, there are always going to be hurdles and concerns that have to be addressed as well. And so after the break here, we want to explore some of those considerations with you, Dr. Boyd. That'll be coming up in a moment here on The FARMSMART Podcast.

Dusty Weis

This is the FARMSMART Podcast, presented by Nutrien Ag Solutions. I'm Dusty Weis, along with Sally Flis, and we're talking today with Dr. Nathan Boyd, Associate Director of the University of Florida Gulf Coast Research and Education Center. 

And Dr. Boyd you did a great job in the first half of the show laying out some of AI's potential strengths in an agricultural application, but what are the problems that it's not going to solve for us?

Nathan Boyd

Well, artificial intelligence is just a tool, and it's one tool in the toolbox. So you should never think that it's going to solve all the problems. It has its own set of weaknesses and issues that, you know, and limitations really, that you can't expect that to do everything. 

Artificial intelligence enables you to automate and extract data, but there are limitations, such as the need to transfer large volumes of data in very rural areas, and that can be an issue.

There's issues with being able to take such a huge volume of data and come out with something that a grower actually can respond to. You can create all kinds of maps, but if those maps don't give you a meaningful answer, then they don't really serve a purpose.

Sally Flis

So as with any technology, there's always lots of stories about when the technology isn't working. I think I actually just saw one in the last week or so where they were demonstrating a robot with AI capabilities, and the robot got bored, the AI got bored and started doing something else instead of what it was supposed to be doing during the demo.

So what are some of the things that we think about using AI in something like our food supply, where we really want to be careful with what we're doing and what's reaching consumers, what are some of the guardrails we need to think about on AI for use in agriculture?

Nathan Boyd

There's a lot of things that you have to be aware of. So for example, the ownership of the data, making sure the growers continue to own the data that comes from their farm and cybersecurity around that data, that, those are key issues. 

Making sure that you understand the parameters within which a decision is made. Because AI is making a decision based on what you feed it. So you have to make sure those parameters fit within what that decision is. 

When it comes to robotics and actuation, so you know, where you're moving to do something. There needs to be multiple sensors that are guiding it. So if you have the failure of one, for example, if you're relying on GPS and you lose your satellite signal, you need to have other types of sensors that can step in, if you can think of it that way, and continue the equipment using the way it's supposed to be. Those are key things when it comes to the construction. 

Things that growers care about are reliability. It's got to make sure that they can rely on this to work when they need it. That it’s not going to stop when it loses GPS signal because, I’m just using that as an example, but those are big issues. 

It's got to be designed in a way that they know how to operate it, but also they know how to maintain it. Because if something goes wrong on a field, you need to have a way to get things up and running quickly.

Sally Flis

Nathan, you just mentioned one in that list of examples that comes up a lot as we talked to Crop Consultants of, how can we use, you know, there's always discussions of how is AI going to do everything, replace a Crop Consultant and give recommendations, that sort of thing. 

And that point that you made about understanding those local conditions and the history and all of the things that come, that cultural experience, from a Crop Consultant, from a grower, to interpret that item you're getting from AI as a recommendation.

It's really interesting to hear you make that comment, because it's definitely a point of concern that comes up from our field people.

Nathan Boyd

Yeah, and when we talk about automation, sometimes people, you know, think we're losing jobs. And I'm going to cover a couple of examples. One thing to think about is there's been lots of examples where automation actually increased the number of jobs because the industry expanded. So it doesn't necessarily mean a reduction in jobs. 

The second thing is when we talk about tools to improve scouting, that doesn't mean that you're getting rid of the people doing that scouting. It actually means they can be more efficient. They can gain more data to provide that input. 

And it also, you know, in the past so much technology was developed for larger farms. So there was this constant pressure, get bigger, get bigger, get bigger to afford the bigger and bigger equipment. 

All of the sudden with AI, we had the capability of saying, wait a minute, we can make a small one that a small grower can use, and then the big grower can buy ten of them, and we can train them to function together as a single unit.

So there's this capability to put tools into the hands of scouts to make them better, more efficient, provide better recommendations. You also have all the sudden the capabilities to use AI to help the small farmer just as much as a big farmer, and that's something we really couldn't do in the past. So that's kind of an exciting way to think about it.

Dusty Weis

It's actually a really remarkable notion, this idea that you can take technology and insights that maybe were only available to industrial scale agriculture before and make those same sorts of insights available to small family farms. I think that that's something that everybody could get excited about. 

But of course, we alluded to it a little bit earlier in agriculture, growers aren't always just going to jump into a new technology because it's new. They want to see a value there. They want to know what's the ROI there. So in plainest possible terms, what is the ROI? What's the value for growers in adopting this artificial intelligence technology?

Nathan Boyd

The ROI is going to depend in every case upon the specific technology, and it's also going to depend upon the size. So right now you can, there's everything, there are AI enabled apps that are free. Well there the ROI is immediate, there’s no cost. 

Then there's weed management equipment that costs in excess of $1 million. And growers are still buying some of that equipment because they see a return in investment and a viable, you know, fit within their production system.

So, it's going to depend quite dramatically on the type of cropping system and the type of technology that you're developing. So it's hard to put an exact ROI. And it's very complicated too, because you can't develop equipment just that cuts back on pesticide use, for example, because if the industry wide cuts back on pesticide use, the pesticide company still has to make money.

So what's going to happen is the cost of pesticide is going to go up, right? So you have to think about what is the benefit besides just a cut in production input costs. Are there other benefits? And there's a lot of them. So one of the, a big issue in agriculture right now is how the EPA and the Endangered Species Act, there are some guidelines that are affecting how growers can use pesticides.

Well, this type of targeting technology, as an example, can give them the points they need to fit the needs of that program. So there's lots, that's just one example, there's many, many benefits of the technology beyond just cutting back on inputs.

Sally Flis

Nathan, I want to just dig into that pesticide example a little bit more. I mean, we hear that all the time on the sustainability side of, well, we can just eliminate this product or we can just keep using less and we're going to get the environmental or sustainable outcome that we're looking for. But you mentioned the pricing aspect from a production side of things.

But when you think about pesticides or any other product use, whether it's fertilizer or biologicals, if we don't think about all the pieces that are impacting those use changes, that cultural piece, how is weed pressure changing? Just using less of something is generally not going to be the answer to get us to all the outcomes that we want to see at the end of the day.

Nathan Boyd

Yeah, I'm a firm believer that all of this AI technology has to fit into an overall integrated pest management system. If we're talking about things that involve pesticides, we can't just rely on a single tool and we should not do that. So we found in my program we develop targeted weed management technologies. We found that we can manage the weeds just as effectively as broadcasting with a targeted approach.

And that's great. But we still need to have herbicides that we can use in the crops that we're using it. If it's a herbicide based system, and you still don't want to rely solely on herbicides because you still end up with resistance, because you're still using that herbicide to control that weed. So you cannot think of this technology in isolation.

It has to be designed to fit into a current production system, and it has to be used as part of an overall program.

Sally Flis

So Dusty, I'm sure I've said “data” on all 57 previous episodes of our podcast. So we're going to go into data again a little bit on this. How much data is enough data to employ something like AI for your data analysis? 

And what are the data points that a grower should be thinking about collecting and storing, or even an ag retail operation be thinking about collecting and storing, in order to start getting some value out of these AI tools that are becoming available?

Nathan Boyd

It depends upon your desired outcome. So if you're, let me give a couple examples, if you're using a real time targeted spray system, you really don't need to collect data per se because it's just doing it as it moves through the field. However, why would you do that if you can collect that data and use it for, you know, accomplishing other tasks at the same time? 

But within that, so you have to think about, okay, if I'm collecting, let's use an example on collecting, just, you know, a red, green, blue camera and I'm collecting images. I can run those images through AI programs and detect all kinds of targets of interest. So I'm using that same data for multiple purposes to create types of maps or look for anomalies or look for problems.

But then it becomes a question, you know, to take 4K video is lot of data, if you're doing it over 100 acres, that's a huge amount of data. So there are techniques you can use, such as capturing snapshots based on the speed of your tractor so you still cover the whole field, but it uses a fraction of the amount of data. That's one way to do it. 

But there's also maybe you don't need to see every part of your field depending on what you're trying to do. So for example, if you're trying to map a particular thing in your field, it's kind of like you can take a quadratic approach where you're taking snapshots that are spaced at, you know, a specific time frame.

And I'm talking all about imagery at the moment. But all of those types of things you can modify to change the amount of data you need. And it really depends on what you're trying to find and what you're trying to do. And that applies to almost everything that comes with AI, even when it comes to using AI for mechanization or to reduce or try to automate tasks, you don't necessarily have to save all that data to do that task.

Sometimes you're only extracting very key points that are critical for you to make a decision for future activities, or to improve the efficiency of the machine itself.

Dusty Weis

Well, and Dr. Boyd, I think another thing that's really interesting about it is, you know, maybe as a grower, I'm not planning to implement an AI solution on my farm tomorrow or next year, but if it's something that I think I might do in the next 10 to 15 years, I should be collecting that data now. Because the more data that we have to plug in, the more effective the artificial intelligence is going to be.

And so, Sally, I think that's why it's really interesting from where we're sitting in that we've been saying for years, you've got to be collecting that data now on your operation. And so to you, Sally, how does what Dr. Boyd is talking about here today and what they're working on at the Center for Applied Artificial Intelligence, how does that parallel with how we're approaching AI from Nutrien Ag Solutions’ point of view here?

Sally Flis

Really well Dusty, we've had very similar conversations. How do we use all the data we have across our multiple platforms to help our Crop Consultants and growers in the field make recommendations faster and have more data going into those recommendations that they're making for performance of a product, for environmental impact, for whatever it is that they're looking to do on that acre.

So it's been really nice to have this discussion and hear about how those things are really starting to line up across the industry, and hopefully we can get away from these conversations about being that fear around AI, replacing people or making crop recommendations. 

Because it really sounds like we're moving in the same direction around what the value really is, which is taking all this information we have across the industry to really refine and make the best recommendations we can make on the ground, and give those to the people who have that cultural knowledge to refine them and implement them in the best way.

Dusty Weis

You know, Dr. Boyd, when we talk about that ten year horizon in the field of agriculture, how do you think artificial intelligence is going to change sustainable agriculture as we know it?

Nathan Boyd

AI has the potential to give us information, especially over time, to track progress or management of issues. And that is really important when we want to make decisions, because often we if we try to make decisions within a single snapshot, it's not looking at: is the problem getting better or is it getting worse? 

Is our approach to a particular problem, is it making the problem better over time, or are we just putting enough input in that we have, we’re at the same place every year? And with AI we can really effectively track over time what's happening in terms of, you know, whether it's nutrient or whether it's disease or whether it's water, all kinds of issues that you can track very effectively.

Dusty Weis

Well, I've got to say from my perspective, as someone who likes to nerd out about these kinds of things, this has been one of the conversations that we've gotten to have where we get to really get down in the nerdy weeds, so I've enjoyed it immensely, and I can tell from Sally's grin that she has as well. 

This has been a fun one, and it's exciting times in the field of agriculture to see what this new technology is going to mean for us here.

So Dr. Nathan Boyd, Associate Director of the University of Florida Gulf Coast Research and Education Center, thank you so much for joining us on this episode of The FARMSMART Podcast.

That is going to conclude this episode of The FARMSMART Podcast. New episodes arrive every month, so make sure you subscribe to The FARMSMART Podcast in your favorite app and visit NutrienAgSolutions.com/FARMSMART to learn more. 

The FARMSMART Podcast is brought to you by Nutrien Ag Solutions, with editing by Matt Covarrubias, and the FARMSMART Podcast is produced by PodCamp Media, branded podcast production for businesses, PodcampMedia.com.

I'm Dusty Weis, for Nutrien Solutions, thanks for listening.