Watch John Deere’s Doug Sauder on Infinia ML’s Machine Meets World.

John Deere’s Doug Sauder on AI Lessons from the (Literal) Field

Join Machine Meets World, Infinia ML’s ongoing conversation about AI

James Kotecki
Sep 8 · 19 min read

Episode Highlights

This week’s guest is Doug Sauder, Director, Digital Product Management & Analytics at John Deere.

“It’s easy to get focused on the exciting algorithm that’s going to be developed, the predictive model that’s going to be developed, but it’s the unglamorous work of collecting data, of assessing data quality, of building data pipelines and robust structures that allow for data scientists to get at that data,” he says.

“Often you’ll find that data scientists will spend 70% of their time just wrangling the data to get it into a useful form. And so those investments in that foundational data acquisition and transformation pipelines, that’s really where the initial focus of a company should start. Because if you don’t have that data, you’re going to really struggle to create value on top of it.”

“Sometimes the conversation about automation gets into, oh, are jobs going to be replaced by robots, those type of conversations. In farming . . . there’s a real labor shortage globally, a shortage in skilled labor. Farmers, our customers, are asking us for more automation. They want the ability for a lower skilled operator to be able to operate a piece of equipment that used to require someone with many years of training.”

“And in addition to that, we’re really talking about automation doing for a farmer what they just can’t do without the technology. And so maybe to, I like to say that we’re helping farmers be better micromanagers. Micromanaging is also a bad word in business, but in farming, it’s a great word. . . . “

“Picture that your job as a farmer is to care for millions of [personal] gardens in a given season. You just can’t do that without technology that can automate and give the precise application of nutrients, the precise placement of seeds.”

“Historically, the most important sensor has been the farmer’s eyes themselves as they observe the physical environment. And what we’re doing with camera technology is really augmenting those human eyes with cameras. We’re augmenting and supporting the human brain with computers, and then we’re augmenting the human hand with robotics. And so those things really come together to give farmers additional tools.”

“This is no different than the type of innovation that we’ve been doing for 180 years. It’s just different than the steel plow. Now we’re talking about cutting edge technologies like AI, but it’s all about helping farmers be more productive, more profitable, more sustainable.”

Watch the show above. You can also hear Machine Meets World as a podcast, join the email list, and contact the show.

Image for post
Image for post
Photo by Loren King on Unsplash

Audio + Transcript

James Kotecki:
Live from Infinia ML, this is Machine Meets World. I am James Kotecki, so excited to talk artificial intelligence today with my guest, Doug Sauder. He is, and I will read this to get it right, the Director, Digital Product Management and Analytics at John Deere. Welcome, Doug.

Doug Sauder:
Thank you, James. Great to be here.

James Kotecki:
Doug, when people hear your title and then they hear the company that you work for, what do they usually think?

Doug Sauder:
Well, James, I’m based in the Bay area. And so when I talk to people and tell them I work for John Deere, they’re a little bit surprised at first wondering, what is John Deere doing in Silicon Valley? But I find that people have a significant interest in agriculture. Many of us have family roots that go back several generations, are familiar with the brand, and I’m always excited to talk about what technology is happening in agriculture and how we at John Deere are not really just an equipment company, but actually a technology company that’s applying technologies like artificial intelligence to solve some of the world’s biggest challenges.

James Kotecki:
And that brand is so strong. When I told my grandfather I was doing this, we had memories of me riding around on the John Deere lawn mower that he had at his house, a kind of riding tractor. One of my colleagues today was really excited for this interview because as you said, he has a family farm and he grew up with this technology. Give me a sense, just to kind of set the playing field, give me a sense of where the technology is today. I mean, this is an AI show, but you can expand a little bit further than that if you want. Like, when we think about the cutting edge of agriculture, what are we even talking about?

Doug Sauder:
There’re some really big challenges that we face in agriculture. As many of you know, the world’s population is projected to grow. And so that creates a challenge for farmers to meet the needs of food production in a sustainable way. And farmers globally face significant challenges of shortages of skilled labor, unpredictable weather. And there’s always a race against time to get crops in on time. And so what this results in is a huge need for farmers to be able to do more with less, and that’s less time, less labor, less seeds, less chemicals, and really technology provides an opportunity to help farmers solve those challenges.

Doug Sauder:
And what we’re really focused on is bringing that technology to bear to help farmers optimize every aspect of their farming operation.

James Kotecki:
And so when we talk about AI and we zoom in on AI specifically, what are some of the things that, I don’t know, what are some of the things that you’re working on right now, this week, what’s going on?

Doug Sauder:
Well, one of the most exciting technologies that we’re developing is called See & Spray. Traditionally, when farmers have a problem with weeds in their field, they have to make a field level decision about whether they should take a machine called a sprayer in and spray that field to take care of the weed problem. And a field might be 100 football fields, just to give you a sense of the size. So it’s a field level decision. What AI enables us to do is mount cameras on that machine. And as that machine goes to the field in real time, those cameras are taking pictures of what’s beneath it. It is running that through a machine learning algorithm that has been trained on hundreds of thousands of images and can detect the difference between a weed and a crop.

Doug Sauder:
And then that algorithm enables individual nozzles to spray chemicals only on the weeds and not on the crop. And so this is an application of AI that can result in reduction of herbicides up to 80%. And so that has a significant impact, not just on the farmers bottom line, but also on the environment as well. Something we’re really excited about.

James Kotecki:
And I think, speaking of it in bottom line terms is very interesting. I’m sure every farmer wants to do the right thing for the planet and as much as they can, as far as not spraying as much of this stuff. But when we talk about an 80% reduction in herbicide, even from what you’ve already said, I think you’re getting the picture that like for me, and probably for many people in AI, we’re kind of city slickers and we need these basic farming concepts explained to us before we can get into this. But it’s fascinating because when you talk about an 80% reduction, let’s say in herbicide, you’re also talking about a relatively expensive process.

James Kotecki:
AI, machine learning, data science, computation, those things aren’t cheap, right? So how do you weigh these things out and figure out, because for me that sounds like a massive expense to get to the point where every individual nozzle is spraying based on a machine learning algorithm trained on millions of pictures. That’s a big process.

Doug Sauder:
It is. At the end of the day, our vision is to help our customers be the world’s most profitable and sustainable farmers. And so we want that technology to achieve a payback for them so that it makes sense to deploy that technology on their field. And so we really see financial sustainability and environmental sustainability going together because farmers want to leave a legacy for the next generation. Many farmers are family farmers. They want to pass down not only healthy farms, but a profitable business to the next generation. So we see those two going together.

James Kotecki:
Are we at a break even point now or have we already surpassed it where it makes sense for those who have the capital to invest in machinery like this, or are there still investments being made and the groundwork being laid for some kind of future break even point where it just, kind of like with electric cars, people talk about in a few years, it’s going to make financial sense just in baseline terms for everybody to get that. Are we there yet with the kind of AI technology you’re talking about with farm equipment?

Doug Sauder:
We are, that’s one of the things that’s exciting about working at John Deere is we’re not just talking about future concepts. Of course, there are more concepts in the pipeline, but we have AI technology deployed in production, in fields today where farmers are making decisions to purchase that technology because of the benefit that it gives them today. So we’re already there.

James Kotecki:
And I think there’s interesting lessons with so many of the different aspects of what we’re talking about, but let’s take the idea of what it has meant and what it does mean to be a farmer, right? When, again, city slickers like me hear the term farmer, certain things come to mind, a kind of laid back easy life on the farm, which I know isn’t really true, but I’m just giving some general perceptions that people might have. And now you’re talking about all of this technology, and I think this mirrors a trend in other industries where workers who maybe previously did not have to be as advanced in their understanding of AI, machine learning, data science, technology.

James Kotecki:
Farmers may not be the ones programming it, but they at least need a passing familiarity with these concepts to be able to get the most use out of this technology I assume. So how is technology changing what it means to be a farmer?

Doug Sauder:
I think of it as another tool in the toolkit for farmers, and we’re really augmenting and supporting farmers and what they do already. So I like to ask people sometimes, what’s the most important sensor on a farm? And historically, the most important sensor has been the farmer’s eyes themselves as they observe the physical environment. And what we’re doing with camera technology is really augmenting those human eyes with cameras. We’re augmenting and supporting the human brain with computers, and then we’re augmenting the human hand with robotics. And so those things really come together to give farmers additional tools.

Doug Sauder:
And really we’ve been, this is no different than the type of innovation that we’ve been doing for 180 years. It’s just different than the steel plow. Now we’re talking about cutting edge technologies like AI, but it’s all about helping farmers be more productive, more profitable, more sustainable.

James Kotecki:
Are there lessons here about human intuition? Because people always wonder, like, what’s the role of humans in this? And we often do talk about augmentation rather than replacement or full automation in situations like this. But do situations still come up where the farmer’s eyes should probably trump what the camera is theoretically telling the farmer do because the farmer just has more intuition or maybe more data that the machine doesn’t have?

Doug Sauder:
I think there’s a learning curve that goes along with some of these technologies. And one of the things that we think about is not just how to build automation, but what is the user experience that needs to be part of that technology to help build the trust and confidence in the automation so that the, whether it’s the operator in a piece of machinery understands, why is this equipment doing something maybe that’s different than I would be doing right now? And of course, there can be cases where we continue to refine the technology. But in many cases, if you present the data, if you present a visualization for the farmer to understand why did the equipment do what it did, it builds and grows their confidence that the automation is really capable.

James Kotecki:
And so how much of an adjustment are these technologies? I know that farmers have been really embracing technology since, you said the early 1800s, at least since before John Deere even existed, there was, metal plows were better than digging with your hands. Right? So technology and farming have gone hand in hand throughout the entire history of the enterprise, but is there an adjustment period, or is this kind of to a farmer, one more incremental addition of technology in a never ending addition of technology that they’re deploying on their farms?

Doug Sauder:
Yeah. I think it’s an evolution. We’ve been building self-driving tractors for the last 20 years, right? So GPS guided equipment and those type of technologies control systems have been around for a while. And I think what we just see now is the transition of sensors, different types of sensors, different types of control algorithms, and different forms of digital software that helps support maybe what was before a manual spreadsheet or handwritten notes and those types of things. But it is the pace of that innovation is accelerating, but I do see it as an evolution that’s continued over time.

James Kotecki:
The reason I’m probing with some of these questions around how people feel when they get these technologies is I think there’s interesting parallels to other industries where someone comes in and they say, “Hey, we’re going to add machine learning to this.” And workers may be rightfully or wrongfully start getting scared or nervous or wondering how they’re going to have to transition. Do you think farming is a good example to other industries, or are there unique things about farming where you mentioned there’s maybe a labor shortage, there’s certain environmental and population constraints or factors here that make it a unique place where, is farming an industry that is uniquely interested in machine learning? Or do you think it has lessons to teach other industries?

Doug Sauder:
I think there are lessons to teach other industries. Sometimes the conversation about automation gets into, oh, are jobs going to be replaced by robots, those type of conversations. In farming, as I mentioned, there’s a real labor shortage globally, a shortage in skilled labor. And really, so farmers, our customers are asking us for more automation. They want the ability for a lower skilled operator to be able to operate a piece of equipment that used to require someone with many years of training. And in addition to that, we’re really talking about automation doing for a farmer what they just can’t do without the technology. And so maybe to, I like to say that we’re helping farmers be better micromanagers.

Doug Sauder:
Micromanaging is also a bad word in business, but in farming, it’s a great word. And if you just picture, maybe if you have a garden at home, you might picture the care that goes into planting a garden and preparing the soil carefully, planting every seed and spacing it out just right, putting just the right amount of water, putting just the right amount of fertilizer on each and every seed. And now picture that your job as a farmer is to care for millions of those gardens in a given season. You just can’t do that without technology that can automate and give the precise application of nutrients, the precise placement of seeds.

Doug Sauder:
And so when you think about it from that perspective, our customers are asking for us to innovate in these areas. And I think there’s probably a lot of application to those principles to other industries as well.

James Kotecki:
John Deere, as a company, as a brand has been around since 1837. Are there unique benefits and or obstacles with the work that you do in data science, in analytics and trying to make that more of a part of your offering? And I assume probably more a part of the internal runnings of the business as well. There’s many large companies out there, many large brands trying to make the shift. What can John Deere teach them?

Doug Sauder:
Well, we’re still on the journey as is every company. I do think we’ve learned a couple of things along the way. We’ve had teams of people building products for farmers for a long, long time as you mentioned. As we think about AI driven products, data science driven products, it’s really, I think I become more and more convinced that it’s a team sport and that the players on the team are maybe different than just traditional software engineering products of the past. If I think about some of the unique roles that we’re bringing together, bringing together data scientists, taking folks that maybe in the past were really academically oriented, focused on their output as a white paper.

Doug Sauder:
Now they’re contributing to production software code that’s running farmers’ operations and enabling their livelihood. Those data scientists really need data engineers to support them to build the data pipelines that bring data from operational systems maybe that collect that data and get it into structures where the data scientists can create models. That’s a relatively new discipline. And then product managers and UX designers, when you’re building digital software tools that are maybe showing the output of a predictive model or an estimate, it’s different than just performing the same calculation over and over again.

Doug Sauder:
And so what’s unique is bringing those new disciplines together and putting them in an environment that are iterating rapidly, working with customers to make sure that software is working well. So I see it as an extension, an expansion of traditional software development. And we’ve learned that that teaming model of bringing those unique disciplines together is really important. And I think the second thing maybe I would say is that your AI strategy has to run on data. It’s easy to get focused on the exciting algorithm that’s going to be developed, the predictive model that’s going to be developed, but it’s the unglamorous work of collecting data, of assessing data quality, of building data pipelines and robust structures that allow for data scientists to get at that data.

Doug Sauder:
Often you’ll find that data scientists will spend 70% of their time just wrangling the data to get it into a useful form. And so those investments in that foundational data acquisition and transformation pipelines, that’s really where the initial focus of a company should start. Because if you don’t have that data, you’re going to really struggle to create value on top of it. So those have been a couple of the lessons I think we’ve learned along the way.

James Kotecki:
And had John Deere been collecting this data before AI and machine learning were major pushes? Is there a benefit there because of course, the company’s been around for a long time, but was it uniquely situated in its data because of just business practices that it had in the past several decades?

Doug Sauder:
To some degree, we have been connecting equipment to the cloud for many, many years, really for the benefit of our farmers. And so one example is the type of diagnostic machine data that comes off of our equipment that has flown up to the cloud for troubleshooting and diagnosing of equipment issues. We were able to leverage that data to create predictive diagnostics. So we now process that machine data. We create what we call expert alerts. So this is a predictive model that says, “Hey, you’re going to experience a part failure in the near future.” We send that message then to our dealers so they can proactively reach out to our customers and service that vehicle before there’s any downtime.

Doug Sauder:
And this is a really big deal because in the heat of the battle, the last thing that a farmer needs is to lose a few hours or a day of work because of a part that went down. So that’s an example where much of that data had existed for similar purposes and we extended its use by use of a machine learning algorithm to create more value with it.

James Kotecki:
Yeah. This is so relevant. And it occurs to me that like when John Deere talks about deploying AI, and ML, and technology into the field, like you’re one of the few companies where that is literally true. And so it’s really exciting to talk to you about this and to hear that so many of the things that you’re going through are things that we hear at Infinia ML from other clients in so many different industries, but that you’re putting it to practical use. That’s great. Are there ethical issue, people talk about ethics a lot these days, obviously bias and facial recognition, and hiring are certain ones, medical issues.

James Kotecki:
Are there a set of ethical issues in precision agriculture and the kind of machine learning AI work that you do?

Doug Sauder:
Well, as you mentioned, when people talk about ethics and AI, they’re often talking about bias and models. It’s something that is maybe a little different for us, but is still relevant. We have a global customer base and that equipment, if we think about an application of AI in equipment automation, we want that to work in all of the physical environments that that equipment is going to operate in. So having really good test sets of data that are broadly representative of the environment that our equipment will operate in is certainly something that we focus on. I think another topic in agriculture that’s really relevant is the use of data, data privacy, and how data use policies are created.

Doug Sauder:
This is something we’ve spent a significant amount of time in, our customers have entrusted us for years. We have a long standing brand and they’ve entrusted us to store and manage data that comes off of their equipments, data that’s used to manage their farm. And so we’ve gone to great lengths to have transparent data policies. In fact, we’ve received industry certification that validates that we’re in compliance with those transparent policies for how data can be used. So those are a couple of the issues that come to mind.

James Kotecki:
Another very timely issue, of course, COVID-19 and agriculture has been a large part of this story. Are we able to supply everybody with the food that we need to get through the pandemic in very difficult times? So when you see the history of agricultural AI being written, what does the COVID-19 chapter say?

Doug Sauder:
I think that the COVID-19 chapter really cements the role of connectivity in particular for helping us get through this pandemic. Of course, most farmers aren’t able to just work from home like you and I are today, but when our customers have a challenge, because we’ve invested in connectivity over the last years, our dealers are able to remotely diagnose and service vehicles, apply over the air software updates using those kinds of remote connectivity tools. And that allows both farmers and our dealer service technicians to do their jobs safely. The ability for farm managers to, from their smartphone, manage their farm, make decisions, oversee what their operators are doing in the cab without having to be there physically allows them, again, to socially distance and be safe.

Doug Sauder:
And I already gave the example of the expert alerts that even then take that to the next level by applying machine learning algorithms to use that connectivity and data to proactively diagnose potential problems. So all of those areas are ways in which, things we were doing before, but the increase in utilization of these technologies that we’ve seen over the last six months has been significant. And we do not expect those to decline even as the pandemic hopefully recedes here in the near future.

James Kotecki:
Are there works of fiction that inspire you for this? I think about my history, the history that I have had personally of fictional futuristic agriculture probably starts at like the old Horizons ride at Epcot, which is no longer around, but it used to show you like futuristic farm robots going around. Luke Skywalker and his family were moisture farmers. The movie, Interstellar, came out a few years ago, the Christopher Nolan movie. The whole first act of that movie was about farming. And they were using a ton of automated robotic machinery to get that done. So are there fictional works that you go to that inspire the work that you do?

Doug Sauder:
Well, one movie that comes to mind, it doesn’t specifically relate to agriculture, but it’s Apollo 13. It really gets at the heart of necessity being the mother of invention and the innovation that was required to bring back the group of astronauts. My favorite scene in that movie is when they dump all those parts on the table and-

James Kotecki:
I knew you were going to say that.

Doug Sauder:
… all the parts on the table and say, “This is what we’ve got to work with.” I think it was to fix the air scrubber or something like that. And so that’s a little bit, perhaps not as dramatic, that’s a little bit how I think about product development though. We’ve got technologies that exist today that are on the table. How can we take those? How can we learn from lessons from other industries, take what’s available, and really use those to assemble great solutions to customer problems and help farmers be more profitable, be more sustainable? And so that inspires me to be innovative every day.

James Kotecki:
Doug, you’ve been an exceptional guest on Machine Meets World, that I want to give you the last minute or two to say anything you want. Is there anything that you want to plug or any concept or topic that we didn’t broach here that you think people should know about?

Doug Sauder:
I think we’ve covered it well. I would just maybe summarize by saying that we’re really proud to support farmers globally in the work that they do. And we’re excited about how technology, including artificial intelligence can really make a big difference, not just in the lives of farmers, but really in the lives of all of us. We depend on a safe and reliable food supply to put food on the table for all of our families. And we think that the future of doing that in light of a rising world population is really the application of artificial intelligence and technology like that.

Doug Sauder:
We’re excited to be leading the way, and thank you for the chance to talk about it today.

James Kotecki:
You summed it up well. John Deere’s Doug Sauder. Thank you so much for joining us today on Machine Meets Worlds.

Doug Sauder:
Thanks.

James Kotecki:
And thank you so much for watching and or listening. Please share, like, comment, et cetera, et cetera, et cetera. And you can also email the show. If you want to email the show, it’s mmw, Machine Meets World, mmw@infiniaml.com. I have been your host, James Kotecki, and that is what happens when machine meets world.

Image for post
Image for post

Originally published at https://infiniaml.com on September 8, 2020.

Machine Meets World from Infinia ML

Weekly Interviews with AI Leaders

Medium is an open platform where 170 million readers come to find insightful and dynamic thinking. Here, expert and undiscovered voices alike dive into the heart of any topic and bring new ideas to the surface. Learn more

Follow the writers, publications, and topics that matter to you, and you’ll see them on your homepage and in your inbox. Explore

If you have a story to tell, knowledge to share, or a perspective to offer — welcome home. It’s easy and free to post your thinking on any topic. Write on Medium

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store