Ep. 37: Artificial Intelligence in Agriculture with Scott Shearer
What is your first thought when you hear “artificial intelligence (AI)?” For some it might be an interesting or exciting concept full of possibilities. For others it might feel a little frightening or unrelatable. In Episode 37 of Precision Points, we wrap up our two-part discussion with Dr. Scott Shearer from Ohio State University (OSU) and focus on AI. Scott walks us through some of the exciting possibilities, helps us better understand the impact AI can have on every operation and encourages listeners to not be afraid of the technology, but to embrace its potential.
Earlier this year the National Science Foundation announced that they will be establishing new artificial intelligence research institutes, adding 11 to the initial seven established in 2020. With this addition, the combined investment is $220 million and will involve two projects at OSU. One of the projects has some agricultural components that Scott is involved with, called ICICLE.
“The thought behind ICICLE is to democratize artificial intelligence,” Scott started. “And right now, as a lot of people might guess, some of the big tech companies rely pretty heavily on the use of artificial intelligence to drive a lot of what they're doing. And so the question is: how do you make it available to everybody to use for everything?”
When it comes to agricultural impact, this project is going to look at digital agriculture. In general, Scott said he feels we will see AI creep into a lot of different elements of agriculture. Specifically, he thinks of digital agriculture through two lenses: crop care and water management.
When considering crop care, Scott walks us through several real applications where we could be seeing the impact of AI. The first he mentioned is John Deere and their See & Spray technology. This technology uses AI to determine what biomass in your field is a weed and selectively sprays it as it goes through the field at field speeds. They have reported, according to Scott, up to a 70% reduction in herbicide application, which can have both financial and environmental impacts.
Scott then explained how OSU is using AI to enhance efficiency of field scouts. They have been using cameras, dropped down into the canopy via a rod on a multi-rotor drone, collecting images to create a database or library archiving the plant stressors they are trying to identify.
“We've been using artificial intelligence to process those pictures,” shared Scott. “A phosphorus deficiency might show up as yellowing of the margins of the plant leaves. The nitrogen deficiency might show up as a yellowing of the midriff of those leaves. With artificial intelligence, we're able to process these images and very rapidly determine what that nutrient stress is.”
Choosing the right hybrid or variety for your field can be one of the most complex decisions we make all year, and Scott said AI could have a hand in improving that process as well. AI can help determine the best fit for your soils by using aggregated data to create a better understanding of how each product performs under a variety of soil types and variables. This can be increasingly beneficial as seed companies continue to race for better, higher-yielding products, resulting in more frequent product turnover.
When it comes to water management, Scott shares that they are looking at more than just irrigation. He is seeing more growers add tile drainage controls to their field tile and thinks the impact of this could be huge for Ohio. For growers with relatively flat fields, controlling that tile is fairly straightforward. However, learning how to best control fields with a lot of elevation relief could have a big impact for the grower and the environment.
On the irrigation front, Scott talked about AI’s potential impact with robotic irrigation. Traditionally, when we think of irrigation, we think of big center pivots. Scott sees robotic irrigators as a smaller applicator that can move field to field. This will allow for more flexibility for things like rented land and fields that don’t have a nearby water source.
“With sprayers, we have a boom that reaches out across the field. And we store water on a tank that's kind of on the central part of the sprayer. With some of these robotic irrigation systems, actually what they're going to do is smaller diameter hose, but they're going to lay a hose down on the field,” described Scott. “They're going to irrigate as they move through the field, using a boom. And then when they go to pick that hose up, they're going to irrigate on the way back. The neat thing about some of these robotic irrigation pieces of equipment is they can preplan where that hose gets laid and where it gets picked back up, so the size or shape of the field, the direction of planting, those factors, they're not quite as consequential.”
As we wrapped up our conversation, Scott ended with asking growers to not be afraid of AI. His hope is that growers continue to embrace the technology because the technology is working for them and their operations. As always, I asked him about one technology he is most excited about and he pointed me to two companies that are working on automation in the field. Sabanto Ag, one of his suggestions, will be joining us on the podcast in Episode 39, so be sure to tune in as we continue that conversation. To hear my full conversation with Scott, check out Precision Points Episode 37.
What are your thoughts on artificial intelligence in agriculture? Comment below – we would love your feedback!
Host: Morgan Seger
Morgan Seger grew up on a small farm in northwest Ohio before studying agriculture at The Ohio State University. She spent 10 years working with ag retail – specifically in ag tech – prior to hosting the Precision Points Podcast. She lives and farms in western Ohio with her husband Ben and their four children. Morgan has her own blog, Heart and Soil, where she talks about her experience farming, gardening, and raising her family.
Guest: Scott Shearer received his Ph.D. in agricultural engineering from The Ohio State University (OSU) in 1986. Currently, he serves as Professor and Chair of Food, Agricultural and Biological Engineering at OSU. Prior to 2011 he was Chair of Biosystems and Agricultural Engineering at the University of Kentucky. Highlights of his research career include development of methodologies and controls for metering and spatial applying crop production inputs (seed, fertilizer and pesticides); modeling of agricultural field machinery systems; autonomous multi-vehicle field production systems; strategies for deployment of UAS in agriculture; and analyses of production agriculture data sets. He has lead research supported by over $12M in grants; authored more than 200 technical publications (refereed journal articles, conference proceedings, meeting papers and book chapters); and has made numerous invited presentations at international conferences, professional meetings and farmer forums. Dr. Shearer is a Fellow of the American Society of Agricultural and Biological Engineers.
Host: Morgan Seger
Guests: Scott Shearer
Welcome to Precision Points, an ag tech podcast where we plant seeds of innovation to inspire informed decisions about precision technology and its impact for growers like you. We explore precision ag tools and technology from the soil to the sky with your host Morgan Seger.
Morgan Seger (00:22):
Welcome back to Precision Points, an ag tech podcast from precisionagreviews.com. I'm your host Morgan Seger. And in each episode, we strive to bring you unbiased ag tech information and ideas. Today on the show I'm joined once again by Scott Shearer, the professor and chair of the Department of Food, Agricultural and Biological Engineering at Ohio state. So if you tuned into last episode, he was on and we talked through the Ag Data Coalition and data management, and we took a turn and we started talking about AI.
Morgan Seger (00:54):
So the rest of this conversation, this is part two of my conversation with Scott, is going to be around artificial intelligence and the way it's being used in agriculture today. Now Scott brought up several things that we haven't talked about on the show, yet. So I really appreciate this conversation that we had and I hope that you enjoy it. Here's part two of my conversation with Scott.
Morgan Seger (01:16):
I'm excited about this. Can we shift gears a little bit? I just saw a release from Ohio State saying that the national science foundation announced $220 million in AI research institutes. And I saw that Ohio State's part of that. And I read that you were leading the digital ag portion of that grant. So can you share what that looks like?
Scott Shearer (01:38):
Well, just for information purposes, there were actually two awards made to Ohio State. There were 11 awards total, nationally. Two of those came to Ohio State University. One of those was called Icicle and I'll come back and talk about Icicle because that's the one I'm somewhat involved with. The other one was actually for use of artificial intelligence in terms of developing 6G or the sixth generation of cellular data communications.
Scott Shearer (02:07):
So two very interesting projects. The one that I'm a little bit more familiar with and have been involved in, in some of the development of the proposal was what we call Icicle. The thought behind Icicle is to, the way we would term it is democratize artificial intelligence. And right now, as a lot of people might guess, some of the big tech companies rely pretty heavily on the use of artificial intelligence to drive a lot of what they're doing.
Scott Shearer (02:35):
And so the question is how do you make it available to everybody to use for everything? And that's kind of the thought behind Icicle. When you work with a group of computer scientists that there's a whole vocabulary there that you kind of get lost in, and this might make some sense to you, maybe it doesn't. Basically kind of a tagline for this project is something along the lines of computational intelligence for artificial intelligence and artificial intelligence for computational intelligence.
Scott Shearer (03:04):
So there's a whole lot to do with computer architectures when you begin thinking about artificial intelligence, and then there is the artificial intelligence we think about algorithms. This project has three use cases. One of those use cases is digital agriculture and I think of digital agriculture and I kind of divide it into two big realms. There's crop care, which a lot of farmers are engaged in right now. And really kind of goes back to the origins of precision agriculture. The other one's going to be water management. And again, two big buckets to put things in when we begin thinking about artificial intelligence.
Scott Shearer (03:38):
We're starting to see some products come to market, and I'm going to point to a John Deere's purchase of Blue River. And their See & Spray technology, I believe, which is trademarked, but the thought is to put vision systems on a spray boom, to identify whatever the crop stress or infestation might be in the case of, I think they're working with being able to identify weeds and that kind of follows along the work that Blue River was doing. But then when you spray, spray individual weeds. Deere has been reporting with some of their initial work, a 70% reduction in herbicide application.
Scott Shearer (04:14):
When you begin thinking about artificial intelligence, in this case, the cameras are collecting images and then you're analyzing those images using what we know now as artificial intelligence to make decisions about whether or not you need to spray. I think the unique thing about John Deere is they've been able, reportedly, to make that work on a 120 foot boom going at field speeds. When you go back to what Blue River was doing, they were thinning lettuce crops out in California, and they were doing that at about two to four miles an hour. And so again, it's been interesting to see the transition of the technology, but also with the computational power that we have day on machines, things that maybe five or 10 years ago, we didn't think were going to be possible are now possible because of the cost of the hardware has come down so much.
Morgan Seger (05:03):
Yeah. So when they are reducing the herbicide, they're still seeing like efficacy? They're still managing the weeds just with a lower herbicide rate?
Scott Shearer (05:13):
You can say lower herbicide rate...
Morgan Seger (05:15):
Or the same rate or appropriate rate, but less...
Scott Shearer (05:18):
You got to be careful because you'll have to apply in accordance with label rates. But as you watch the machine go through the field, spray nozzles are turning on and off, much like section control only in this case, they're turning on and off as weeds are encountered in the field. So like I said, there's some pretty good videos out there now where they've actually included some dye in the spray material as they go through the field. So you can see the end result of what's being sprayed and the areas of the field that are not being sprayed.
Scott Shearer (05:52):
But historically what we've done with a lot of our herbicides is they've been broadcast applications. In other words, we cover the entire field and that's still going to continue for a while because in the spring time we get those weed flushes and you just about have to spray everything. But as the growing season goes along and we go back in and we make those subsequent treatments, especially post emerge, that's going to be right for this technology in terms of being able to conserve herbicide.
Scott Shearer (06:19):
And so again, some of the numbers that Deere's reporting, right now, it gets my attention. I think it probably gets farmers' attention. Of course, there's going to be a trade off between the herbicide bill and the technology bill. So let's not forget that.
Morgan Seger (06:34):
Right. Well, it's really interesting because I feel like where I've heard people talking about the application of this type of technology using AI to identify weeds or problems in the field, it's been with smaller autonomous devices. So do you think that there's more of a future going that way or do you think it's more of keeping an operator in the machine, but having the machine work more autonomously when it comes to selecting where it sprays?
Scott Shearer (07:02):
I think there's going to be a couple different avenues when we think about how artificial intelligence is going to be used. Let's not rule out the crop scout and their use of artificial intelligence. One of the things we've been working on at Ohio State is we have a camera that we can insert into the plant canopy using an unmanned aerial system. And so we have this camera head that's kind of on a rod that's suspended below a multirotor unmanned aerial system.
Scott Shearer (07:29):
I'll give an example here where I think this becomes pretty effective. When we look at nitrogen deficiency in a corn plant and if flying over the top of the crop with the drone and we notice the leaves and the top of the canopy of yellowed, we can conclude, well, that's probably a nitrogen deficiency. But what I always tell people is that's kind of a post-mortem, okay?
Scott Shearer (07:50):
In other words, by the time you see that deficiency at the top of the plant, it's going to be hard to go in and do a rescue application of nitrogen and recoup the yield loss that might've already occurred.
Scott Shearer (08:00):
On the other hand, what we've been doing is dropping the camera down into the plant canopy and taking pictures within the canopy. And generally what happens in a corn plant is the first signs of nitrogen deficiency are going to occur in the lower leaves. Vertically, we hover and we drop this camera down into the plant canopy and we take the pictures. Well, we've been using artificial intelligence to process those pictures. And so a phosphorus deficiency might show up as yellowing of the margins of the plant leaves. The nitrogen deficiency might show up as a yellowing of the midriff of those leaves. And so with artificial intelligence, we're able to process these images and very rapidly determine what that nutrient stress is.
Scott Shearer (08:45):
We've been creating image libraries for corn plant canopies. We've been creating them for our soybean canopies as well. Typically, we're looking at about 8-10 stresses that are common in the state of Ohio. And I had a graduate student working on this and he was creating these libraries over multiple years. To give you a feel though, we're not quite at the scope and scale we need to be. The image data sets we've been working on for about 10,000 images, but really we need to be in the hundred thousand to a million image category.
Scott Shearer (09:13):
And so therein kind of lies one of the challenges that we have in agriculture with this evolution of artificial intelligence, and that is creating the classified data sets that we can use to train what we call the classifiers. And that is the intelligence on the machine that's going to make that split-second decision.
Scott Shearer (09:33):
We've been processing our images in the field at the rate of 15 images a second. In other words, as we go through the field and we're collecting these images, we're processing on the drone and making the decision right there. Can we transition that to a sprayer? Well, obviously Deere has been doing something very similar.
Scott Shearer (09:50):
The question is, where's the technology going to reside? Our thought is we want to extend the capability of the crop scout. When the crop scout goes to the field with the drone, hopefully they can cover more acres and they can see more of the ground and then make better recommendations to farmers. We'll probably go to the point where we map some of these stresses and go back in and do a post, I'm going to say sensing flight treatment.
Scott Shearer (10:15):
It's going to get to the point where, and again, what Blue River is doing with John Deere in terms of putting that intelligence on the machine to make those decisions in real time as the machine moves through the field. And so I see a spectrum of use of artificial intelligence. Some of that artificial intelligence too, is going to be used to match crop genetics, or plant genetics with the soil landscape. And that's going to occur from season to season. So you've got a little bit more time to process that data and make those decisions, but we're going to see AI creep into a lot of different elements of agriculture.
Morgan Seger (10:50):
Do you have an example of how matching genetics with field type would work?
Scott Shearer (10:56):
We saw an introduction of the multi-hybrid planters. Great technology. As we go across the field, we can switch from hybrid A to hybrid B. The question though that I think probably occurs to most farmers is, is how do I use that technology to make money. And fields that are very uniform, I don't know that there's going to be a lot of value there. In other words, I'm just going to plant either a hybrid A or hybrid B in the field. And I'm going to take what I get at the end of the season.
Scott Shearer (11:25):
However, we do know that a lot of farmers have fields that are variable in nature. And in some locations in Ohio, we get fields that have a gravel lens in them through the middle of the field and whatever hybrid you plant in that region, you're always a week away from drought if you don't get good rains or whatever. And so we'll see some ridges or some of these gravel lenses where the plant always looks drought stressed. Well, you want to plant a drought tolerant variety in that location. Then maybe other parts of the field have much better top soil depths that are moisture storage capacity. We used to use terms such as racehorse hybrids or whatever. And so I think that's how we begin thinking about how we might match up plant genetics with the soil landscape.
Scott Shearer (12:14):
A word of caution though, when you look at most of the offerings of many of these seed corn companies, you're going to want to have about the same maturity dates on your hybrids. And so seed corn producer A sells 16 hybrids in Ohio. But when you look at 105 day corn, there may only be three or four in that category. And so we're going to have to see a little better evolution in terms of how some of these genetics are made available to farmers and how they're going to be managed within the field. In other words, I think there's potential there. I don't know that we've figured out how to capture that potential and turn it into increased revenue for farmers yet.
Morgan Seger (12:56):
Well, it makes me excited for two reasons. One is, when it comes to creating a plan for a field, it can be really hard with one hybrid. So I know I wrote a blog post earlier this spring about creating a variable rate seeding recommendation, and I ended up making it two posts because I had so much to say and at the end of it, it was like, and that's for one hybrid. So if you want to do two hybrids, you have to kind of rethink all of this.
Morgan Seger (13:22):
But then the other thing is, the hybrids and varieties that we're using seem to rotate so quickly. So if there's a way to collect data that can expedite our knowledge and experience with the hybrid before we have to go place it for the first time, I think that could be really valuable to growers.
Scott Shearer (13:37):
Yeah. And I think Morgan, you make an excellent point. It used to be, universities did yield trials and when a hybrid came out, it was in the marketplace for 8-10 years and a lot of farmers would look at those university yield trials and go "Well, they're kind of an independent third person." Or whatever, but today if we use that same approach, the question is by the time the university publishes that data, is that hybrid even any longer available.
Scott Shearer (14:06):
We know some hybrids stay in the marketplace a little bit longer, but we also know the companies have a pretty high rate of new hybrid introductions every year as well. So I think we're kind of forced into a situation where we need to rely on a lot of the seed corn producers, if you will, in terms of the characteristics of their hybrids as they come out.
Scott Shearer (14:29):
The other thing though is we are starting to see because of aggregation of the data, looking at how some of these hybrids perform over a lot broader spectrum of soil types, as well as environmental conditions, precipitation, things like that. So there's some good things coming along with the aggregated data. And I don't know that we're going to see a reduction in the rate of introduction of new hybrids anytime soon. A lot of great technology out there too, I don't want to minimize that.
Morgan Seger (14:56):
Yeah. So when it comes to using AI for water management, is that mostly going to be directed for people who are using irrigation?
Scott Shearer (15:06):
Yes and no. I want to get people thinking about water management in a slightly different light and I'm going to use the state of Ohio as an example. Tile drainage is essential in a lot of areas in the state of Ohio because our springs can be kind of wet and falls can be wet during harvest. The points that I want to make is, we're beginning to learn about controlled drainage as being of significant value to the farmer and also to the environment. So if you think about things, we probably want to open our tile drains up in the spring and let them run so we can drain the land and we can get in and farm it. And Northwest Ohio is extensively tile drained, is one of those examples. During the growing season, then maybe we want to shut those tile drains off and then we want to conserve water.
Scott Shearer (15:57):
Then when we get close to harvest again, we open those tile drains back up, let them flow and make certain we'd get into the fields and then harvest the crops. The point is in between how do we manage water? And so for relatively flat fields, there's kind of one management approach, but we also have a lot of fields in the state of Ohio that have quite a bit of elevation relief.
Scott Shearer (16:18):
So we're starting to see a lot of control structures put on these tile drainage outlets and we begin opening those up or closing them up depending upon the time of year and whether or not we're trying to conserve water. Obviously during the winter months, we're going to block off that tile flow and try to retain as much of that water on the land as we can. There's some tremendous environmental benefits to doing this. A lot of our offsite movements of nutrients really occur with some of the major rainfall events early in the spring time and throughout other parts of the year as well. But when I began talking about water management, yes, we can talk about irrigation. There's no question about that, but let's not forget on the other end in terms of how we're going to control the drainage or water moving off the land.
Morgan Seger (17:03):
Yeah. And then you're also reaching a large audience of people who don't have irrigation, but are still trying to figure out these water issues.
Scott Shearer (17:11):
That's exactly right. And by the way, right now we see a lot of tile drainage going in because hey, prices look pretty good on grain. The other thing I'm going to mention too is historically, we've always thought about irrigation as kind of being, we go back in time, the hand move, more recently center pivot irrigation. But I'll also mention there's a new type of irrigation coming and that's robotic irrigation. I think it could be a bit of a game changer.
Scott Shearer (17:39):
And my point being is center pivot irrigation is well-suited to certain parts of the country, depending upon water supply. It's capital intensive. Everybody understands that, but generally when you buy a center pivot irrigation system, you put it in one field and it stays there for the life of the system.
Scott Shearer (17:55):
Robotic irrigation is going to allow us more flexibility in terms of being able to move irrigation systems between fields and even between farms. May not be done necessarily during the growing season, but it's going to be one of those systems you can easily move if you're renting land between rented fields between cropping season. So again, interesting time in terms of how technology's going to change how we think about water management.
Morgan Seger (18:23):
Yeah. So I'm envisioning like Wally with a backpack with water on it. Is that what we're talking about here? Are you talking about something that looks more like a pivot, that's going to move through the fields?
Scott Shearer (18:34):
Well, I'm going to talk about something that kind of looks like a combination of some of the things we've seen before, but it's a bit different. With sprayers, we have a boom that reaches out across the field. And we store water on a tank that's kind of on the central part of the sprayer. With some of these robotic irrigation systems, actually what they're going to do is smaller diameter hose, but they're going to lay a hose down on the field. They're going to irrigate as they move through the field, using a boom. And then when they go to pick that hose up, they're going to irrigate on the way back. The neat thing about some of these robotic irrigation pieces of equipment is they can preplan where that hose gets laid and where it gets picked back up.
Scott Shearer (19:18):
So the size or shape of the field, the direction of planting, those factors, they're not quite as consequential. In other words, with some of these newer systems, they're high clearance and they don't knock down corn like some of the more traditional irrigation systems might. I'm thinking about the drag hose systems and some of the big gun irrigation with reels where the hose had to be pulled out and then it was retracted as the irrigation occurs.
Scott Shearer (19:45):
But again, some of these systems you'll be able to set them up in a field up to about 160 acres and you'll walk away from it and you'll supply it with water, but it will irrigate the entire field. So again, a unique opportunity. We have to be careful though, because there is limitations on how much water these systems can deliver during the growing season.
Morgan Seger (20:07):
All right, you got my wheels spinning now because I'm just thinking, the impact we could have on our crop yield. If we could give it water when it needed, it could be just astronomical.
Scott Shearer (20:18):
Kind of the robotic irrigation systems, we're thinking about systems that will apply probably about four inches of water during the growing season. And it might make 12 passes over the field.
Scott Shearer (20:30):
The other thing though, and I think the thing that is most exciting is these systems will be excellent for delivery of nutrient sources during the growing season. And more specifically, animal producers will have more latitude to do apply animal nutrient sources during the cropping season. In other words, you've got an actively growing crop in a lot of cases, and it's not like you're going to go out and do a lot of animal nutrient source application. You wait until after the crop comes off or you'll do it before planting. But with robotic irrigation, I think it's going to change that to more of a year round opportunity for farmers to manage some of those nutrient sources.
Morgan Seger (21:12):
Yeah. So that would help out the livestock farmers and could also have an environmental impact for getting it out there when there's a growing crop that can use it.
Scott Shearer (21:18):
Well, I think you hit the nail on the head. Farmers are probably going to find they're able to better use those nutrients sources in terms of translating those into yield. And you're right, it's also going to have profound impact on the environment.
Morgan Seger (21:33):
So if I can ask a follow up question to your drainage system with controls. I know in my part of the world, I'm in Western Ohio, we don't do a lot of fall applied fertilizer, but in other parts of the Midwest, they do. Do you think that if we're going to move to a situation where we could close the tile over winter, that that would help preserve more of what we're putting out in the fall or does that all just kind of wash out when we open them up in the spring?
Scott Shearer (22:01):
Well, when I think about offsite moving the nutrients, I kind of think of two approaches. For the more mobile nutrients, they'll move down through the soil profile and into the tile drainage system. And obviously blocking off tile drains are going to stop the movement. At least the offsite movement. The other thing is for a lot of our surface applied nutrients, we have to be careful with rainfall events. The bigger rainfall events tend to move some of our fertilizers offsite and a lot of times the fertilizer will be attached to soil particles.
Scott Shearer (22:32):
Both cases are very unfortunate situations because I think most farmers realize I want the nutrients on the field when the plant needs them. And that's the most critical thing. As time goes along, I think we're going to see a greater emphasis and shift towards applying nutrients to the actively growing crop when they're needed.
Scott Shearer (22:53):
I'm not going to say that we're going to move totally away from fall applications or surface applications or anything of the like. But I guess what I'm arguing for in some respects is if everybody does the same thing at the same time, when we get a big rainfall, then the environmental consequences are disastrous. On the other hand, if we can begin to spread the nutrient applications out and better match them to the crop needs, I think that serves the interests of everyone. I don't know a farmer that wants to spend money on fertilizers that's going to run off of his or her land and into a surface water.
Scott Shearer (23:27):
So I think we're moving towards a time and era, when we think about technology, it's really going to serve the interest of the farmers in terms of their profitability, as well as farmers concerned about environmental quality, as well as other people concerned about environmental quality. So technology may be that a silver bullet that everybody's been looking for.
Morgan Seger (23:49):
Yeah. And it's awesome to watch it progress and get closer and closer and solve more difficult problems.
Scott Shearer (23:56):
Yeah. And I want to go back and I want to emphasize, not all technologies are going to be appropriate for all farmers. I think depending upon the nature of the operation, it's going to be up to farmers to select those technologies that make sense for their operations.
Morgan Seger (24:15):
Is there anything else you think our listeners should know or be thinking about when it comes to artificial intelligence today?
Scott Shearer (24:19):
Don't be afraid of it. A lot of artificial intelligence a lot of people tend to treat as a black box. It works, well, why does it work? Well it just does. We're moving into an era too when explainable AI is going to kind of rule the day. In other words, if things don't make sense to people, they're not going to adopt them. And we may still have some problems with artificial intelligence in terms of some of the spurious results that you can get.
Scott Shearer (24:44):
And so as we move forward, people are going to gain confidence with the technology. They'll begin using it because it works. I'm just cautioning farmers, be careful that you don't off-hand dismiss it just because it's something different or something you may not appreciate. You also have to be careful on the other side, be careful not to overly embrace it either because it's still in its infancy in terms of applications in agriculture.
Morgan Seger (25:11):
I really appreciate that because I was going to ask if we should be afraid of it because there's a lot of things that we just don't know yet. I've actually never heard explainable AI as a term before, but I like that a lot. I think having some of that clarity will help people feel more comfortable.
Scott Shearer (25:28):
Whenever I talk to producers about artificial intelligence, I try to liken it to indicate to people that we're trying to give machines the ability to think much like humans. And one of the examples I always use is a child pouring milk. You've probably seen the situations where their first attempt probably is not that efficient. A lot of the milk ends up on the floor, but because of their experiences, they learn over time, they get better motor skills. And then before long, the child is able to pour a glass of milk without spilling anything.
Scott Shearer (26:02):
Well, we're kind of in that mode of being able to teach machines the same way. In other words, their experiences lead to better outcomes in the end, but it's going to take a little while. Technology's great when it works and I think everybody appreciates that, but I know there's also a level of frustration when technology does not work.
Morgan Seger (26:25):
Well. If people are listening and interested in the work you're doing or want to follow along with some of the research, where would you recommend they go?
Scott Shearer (26:33):
There's two or three things we try to do at Ohio State. One of the things that has been the focus in John Fulton, Elizabeth Hawkins, the kind of the people leading this, but we do a lot of on-farm research and we publish that through something we call E-fields. It's an annual publication. The 2021 publication will be expanded over what we have in 2020. And that will be coming out the first week of January is the plans right now.
Scott Shearer (27:00):
So all the data that we're collecting from on-farm field investigation, with farmers in the state of Ohio, that's going to be available that first week of January. And so we try to focus a lot on the application of technology and the use of technology. We don't have a lot of examples of artificial intelligence, but I can assure you that will be coming in future issues.
Scott Shearer (27:20):
Morgan Seger (27:55):
One last question I like to ask our guests on the show is if there's one technology that you're most excited about.
Scott Shearer (28:01):
I've been watching a couple companies and I think it would be very good for farmers to be paying attention. One of those companies is SabantoAg, I think they're out of Chicago, Illinois. This year, they were running several Kubota M5 tractors in fully autonomous modes, and they were seeding soybean crops in the Midwest. They got some pretty good videos on, I think, YouTube videos and I believe that they have a lot of Twitter, they're shorter videos. I encourage people to look at what SabantoAg is doing.
Scott Shearer (28:34):
The term that I would use to describe what they're doing as farming is to service. When you talk to Craig Rupp about what he's selling, he's basically in my estimation selling a service. There's going to be some interesting opportunities coming up in the future, but lightweight equipment. There's an indication that in the Casey's M5 tractors are running shifts of 30 to 40 hours, 24/7 around the clock type operations.
Scott Shearer (29:01):
The other organization, I think people need to pay some attention to is a company by the name of SwarmFarm out of Australia. They've been running autonomous sprayers, probably not unlike what Craig Rupp is doing. I think his business model is a per acre charge in Australia. The last time I talked to the owners of a SwarmFarm, they were leasing their autonomous sprayers. There would be a technician, would come out and set up the field, set up the autonomous sprayers. I think they had 30 horse diesel engines on them, 30 foot spray booms. So not huge machines, pretty small.
Scott Shearer (29:37):
When I was down there a couple of years back and visited, they had five of those in one field, running simultaneously. We did some numbers and again, Australia is a bit different than the Midwest here, but if I'm not mistaken, when I did the conversions that they're charged for that service spray application service, was about a dollar an acre. And so again, I look at those companies and I think about where the future might go or whatever.
Scott Shearer (30:06):
And again, I encourage people to be thinking about this term, 'farming as a service'. In other words, in the future, maybe farmers are not going to own all their equipment, then maybe they're not going to have to manage all of this technology at least independently by themselves. So keep an eye on those companies and where this automation is headed.
Scott Shearer (30:25):
I'll also remind everybody that John Deere just announced they were purchasing Bear Flag Robotics. CNH Industrial announced that they're going to acquire Raven and I'll remind everybody, Raven acquired Smart Ag. I think they acquired all the patents from Jaybridge Robotics. I think they now own DOT as well. So some very dynamic things going on in the machinery industry with automation.
Morgan Seger (30:52):
Yeah, for sure. We have a lot to keep an eye on.
Scott Shearer (30:56):
Well, I look forward to coming back and providing an update at some point in the future on some of the things that we're seeing on the automation front as well.
Morgan Seger (31:04):
Yes, for sure. We'd love to have you back on and I'm so grateful for your time today.
Scott Shearer (31:08):
Thank you very much, Morgan. I enjoyed being able to share some of my thoughts with you today.
Morgan Seger (31:14):
Thank you so much, Scott.
Morgan Seger (31:16):
Thanks again to Scott for being so generous with his time and insights here on the Precision Points podcast, and thanks to you for tuning in to another episode. As always, our show notes will be available at precisionagreviews.com. While you're there, be sure to check out our grower sourced reviews. We're constantly trying to collect information from growers like you, about precision ag tools and services that you have used so we can make more informed decisions about precision ag on our operations. Until next time, this has been Precision Points podcast. Let's grow together.
Thanks for tuning in to today's episode. To hear more podcasts like this, please rate, review and subscribe to Precision Points. Visit precisionagreviews.com for show notes from this episode and read expert advice on the blog, share your experience with the Precision Ag products you use and check out our network of farmer reviews. Let's grow together.