We are entering the era of pervasive intelligence thanks to AI. David Schatsky, leader of Deloitte's trend-sensing program, says it is high time companies took stock of the impact this will have on their businesses and position themselves to reap the benefits.
“For every story about workers being displaced by technology, there are other stories about workers being empowered by technology. And I can’t comment on how those two sides balance out against each other, but [I can say] construction workers don’t work with picks and axes anymore, but there are still lots of construction workers and they use power tools. This technology can be thought of as a power tool to help people work more effectively.”
Tanya Ott: What exactly is the technology he’s talking about? We’re going to break it down in this episode of the Press Room.
I’m Tanya Ott, and in this podcast we talk about some of the biggest issues facing business today. We can talk about technology—but I always feel like I understand it much better if I get to see it in action.
Know more about AI and Cognitive technologies
Listen to more podcasts from Deloitte Insights
Subscribe to iTunes
Subscribe to Deloitte Insights
Download the Deloitte Insights app
Calvin Perry: Sorry if it smells like a work truck.
Tanya Ott: It’s fine!
Calvin Perry: Gets a little sweaty in the summer.
Tanya Ott: A while back I took a trip to Camilla, Georgia, to meet up with Calvin Perry. He and other Ag [agricultural] scientists at the University of Georgia run this research farm where they test new crops and new technologies. We drive down a dirt road, passing a corn field on the left. That is the center of innovation.
Calvin Perry: We have this four-tower center-pivot system fitted with variable rate controls …
Tanya Ott: Let me translate. You’ve probably seen one of these things in a field before. It looks like a giant metal skeleton of a caterpillar. There are segments—this one has four—that are attached to long legs and wheels that can roll around a field. Sprinklers on the spine of the caterpillar—if caterpillars had spines—turn on and off to water the field. There are about 7,000 center pivots in the southwest corner of Georgia. That’s more than half of all center pivots in the state. But this one is special because it’s outfitted with special sensors that help determine how much water the ground gets.
Calvin Perry: Okay, we’re at the pivot point. This is the control panel where the user interfaces with it. You set the amount of application in terms of inches applied based on how fast you walk the center pivot around the circle. The sprinklers simply apply water at a constant rate. You can vary how fast it walks and turn sprinklers on and off. We’ll see that in a few minutes.
Tanya Ott: An Ag specialist named BJ is at the controls.
Calvin Perry: He uploads a map from his desktop software. It shows the whole field. And he has the zones and the rates of water that he wants to apply on that thumb drive. Saves it to a thumb drive, brings it here, uploads it …
Tanya Ott: This well puts out 500 gallons of water a minute, and BJ can selectively turn off specific nozzles on the system as it travels over the field. They use IoT [Internet of Things] sensors to measure and collect temperature, soil conditions, and moisture. Those sensors send that data to the internet and ultimately to the farmer. It’s a far cry from how watering is traditionally done.
David Schatsky: I've been following artificial intelligence at Deloitte almost since I've been here and it is, as you know, an area of really rapid progress. It's not just how powerful the technology is but where it gets used that's really starting to change business and society. When people think of AI now, they think of the internet, social media, software that they use at your desk. But what's happening [is] that AI is really taking root in the products and machines that we see and use all around us. So, finding AI at work in a field on a farm is the new normal. And that's really what this report was about, how that is changing industries as diverse as agriculture, construction, medicine, urban planning, and so forth.
Tanya Ott: What are you seeing in terms of AI technology showing up in the construction world?
David Schatsky: One of the coolest things that I've seen is the use of AI too to help monitor and analyze construction projects while they're in progress. I speak from experience because my 32-square-foot bathroom has been under renovation for three months now (laugh) and construction projects can tend to run long, off schedule, and over budget. Large projects—big buildings and bridges and so forth—are really, really tough to manage and track. One company we saw developed an autonomous wheeled device that can roam freely around construction sites, use computer vision to track the progress of construction, and, by capturing and analyzing images, can figure out what part of the project's done and what part's not done, how much construction material has been used and not yet used. And it can pair that up with analytics to figure out whether the project's on schedule or not. This kind of approach can improve productivity and cut wasted materials, correcting things before they go on too long, if a problem is detected, and reduce costs.
Tanya Ott: I want the visual here on this thing that's deployed in a construction site. Is it almost like one of those robot vacuum cleaner kind of things, but it's got a camera and it's sensing?
David Schatsky: It's like that, but it's cooler because this particular one, it's on wheels. It can climb over piles of things. It can now go upstairs and it's got cameras and sensors on it. It can move about autonomously and move around obstacles and scan and take pictures of the whole setting.
There are other approaches to this that use flying vehicles, drones that capture different kinds of information, but the idea is the same in that there is intelligence embedded in these devices that make it possible to automatically do something that, even if you had an army of inspectors and clipboards running around, they couldn't do it because just to analyze the data that they gathered would be too hard.
Tanya Ott: That's fascinating. You mentioned urban planning. How is AI impacting urban planning?
David Schatsky: So, you know the buzzword “smart cities,” and you always wonder if things are getting smarter or getting weirder, right? We found an interesting example there in traffic control. Traffic lights [were] usually controlled by a central system that was programmed by somebody who did a study and decided what the best sequence of lights [was] and how long they should be green and so forth. Now we've seen some pilots where the traffic lights have their own sensors. They have their own intelligence. They can see when there's traffic, when traffic is light or not. They can communicate with each other and learn patterns and can self-adjust and thereby reduce traffic and wait time.
Tanya Ott: Is that like what I remember hearing about in Singapore where they would do surge pricing on toll roads and things like that—they were able to very immediately sense what the usage is and adjust for it?
David Schatsky: So, some cases, like in London I believe, they have congestion pricing, which is based on the time a day. But when you have sensors and analytics you can do pricing in a way that would be dearest to an economist's heart, which is based on demand from minute to minute. And not just pricing, but optimization [also], so that you can either change the pricing dynamically, which can annoy people, or just change something simple like the pattern of traffic lights to optimize the flow of traffic.
Tanya Ott: Interesting. OK, we've talked construction, we've talked transportation, we've talked agriculture, but it goes well beyond that. It's reaching into manufacturing and health care and energy and security, logistics, and distribution. Any other really compelling examples that would paint a picture for us?
David Schatsky: In medicine there's a couple of examples of different kinds that I really like. One is a device meant to treat epilepsy in patients. This is an implanted device that uses machine learning. It receives and analyzes signals from the patient's brain and has been trained to be able to recognize the signs of an impending epileptic seizure before the seizure happens. It can observe the telltale signs and it can then emit some electric stimulation to the brain to avert the seizure in many cases. So, this is a device that, before anybody could tell, was aware that a seizure was coming and could take preventive action and mitigate it or prevent it.
Tanya Ott: And how widely is this kind of device deployed?
David Schatsky: It's still in clinical trials. There is a company that makes this particular one that raised US$74 million after passing a new threshold in clinical acceptance, but there's probably—according to the company—over a million people in the US whose epilepsy isn't well treated by other means. And this could have a huge impact on their quality of life and on the cost of delivering care to these people because they won't need emergency treatment as much as they did before.
Tanya Ott: And that's the real key. It's about how it can help an individual, but then also help the system. And powering all this are these AI chips, which you write are faster than they've ever been. How much faster and why does that matter?
David Schatsky: What's driving all this, apart from the demand and the opportunity to use it, is that the enabling technology is getting better and better. So, as it's become apparent that AI is so powerful, the makers of chips have been designing chips that are optimized for doing the kinds of calculations that happen in machine learning, and it's critical to be able to do this with low power consumption and fast enough that it's practical to do AI processing on a mobile device. Until recently, any serious AI work was done on a big server somewhere in a big data center, but now you've got powerful AI capabilities on your mobile phone, embedded in a camera, embedded in a farm equipment that’s spreading herbicide, and so forth. And for a device that is implanted in an epilepsy patient's body, for instance, you need something that runs on really low power because there's not a lot of space for battery, not a lot of power available.
Tanya Ott: So, they're running really effectively with a lot less power. They're also, I guess, not connected to the cloud or they don't have to be internet-connected. They're working sort of external to that?
David Schatsky: Exactly. Here's the point—back in the old days you would maybe capture data, send it to the cloud for processing, AI algorithms would run there, and they could calculate a result, and send back their results. Nowadays if the processor that is embedded in a patient or installed in an herbicide-spreading machine or on the construction site is capable of doing all of the analysis itself, then it doesn't have to send data back and forth to the cloud to determine what it's seeing and how to respond to it.
Tanya Ott: And that sounds like that would be really powerful, not only because it maybe in microseconds speeds things up, but also if you've got places that have connectivity issues, this would work around that some.
David Schatsky: Yeah, it makes this kind of stuff available where connectivity is not great and where latency is not good. That is to say, where you need instant results and having to go to the cloud for the answer is not going to work. This kind of thing can be applied in remote oilfields, for instance, monitoring and taking corrective action when equipment malfunctions. Or out of the middle of the farm field. This stuff will definitely find use in autonomous vehicles and cars.
Tanya Ott: We're looking at greater efficiency, greater effectiveness. What is the potential downside?
David Schatsky: There's always a downside. That's my favorite part in some cases. What I'll say is that this technology is going to open up new markets with AI-embedded products that can be used when dumb ones maybe weren't useful. A simple example that explains the point is in surveillance. So back in the day, if you had an area that needed to be protected, you'd put a video camera there and somebody would have to watch it. But a lot of times there's not enough people to watch all the video that's being collected or the area that you want to surveil is far from a place that you could have somebody station to watch things. Now smart cameras cannot just transmit video but can actually tell you when it's time to pay attention because they detected some kind of behavior that is suspicious. So, there will be more surveillance cameras sold where they weren't even useful before.
Another example is in medicine, a company that I was talking to the other day; [it] has developed an AI-powered handheld sonogram device. If you’ve ever had a sonogram, they’re big machines that can cost US$50,000. This is one that can fit in your pocket and is selling for US$2,000 now and the cost is dropping. This is going to reach all kinds of new markets that were off limits to the big machines before, including emergency medical technicians and medical personnel in poor rural areas that could have in no way afforded the big machine before.
New markets is one impact of all this stuff. Another impact, though, is the kind of threat to incumbent makers that are making traditional products that are not smart. But to me, the thing that is most interesting about this is how it's going to impact how money is divided in these industries.
Tanya Ott: What do you mean by that?
David Schatsky: How the profits are allocated. Take the agriculture example.
One of the [examples] we saw was a herbicide spreader used to spread weed killer on farm fields. And we have one manufacturer who's built a machine that uses computer vision to scan the field as the equipment's moving over it and to determine exactly where to apply the herbicide in targeted ways and to not spray it where it's not needed. The implication of this is that a huge amount of herbicide can be saved when you're only spraying where you need it. In some tests they reduced the amount by 90 percent that was needed. And what that means is that somebody who's selling herbicide is going to sell a lot less herbicide.
Tanya Ott: Right.
David Schatsky: Smart machines are going to earn a lot more revenue. So how the revenue gets allocated in an industry is going to be changed by this, because materials will be used much more efficiently. Even in the health care example, there may be fewer emergency room visits, thankfully, for epilepsy patients but there'll be more money spent in the procedure implanting those devices and monitoring them. This is going to play out in construction as well, in [reducing] construction waste. It's going to impact the demand for labor, but it also will lower the cost of building and so it could spur new building developments. The economic implications of this are pretty interesting.
Tanya Ott: I'd be curious to know, let's say in the herbicide example that you have, there is a large company that I won't name that works a lot in herbicides and I'm wondering if that kind of company has been looking at this and saying we're going to lose market or revenue to this new technology so we need to get into this new technology. Or do they just figure that's the tech people and we’re the herbicide people?
David Schatsky: This is the strategic and, in some cases, existential choice that companies will be faced [with]. And there is a story of a company that sold film, and demand for that film dropped precipitously because digital cameras made it unnecessary.
Tanya Ott: So, this potentially means every company has to be a tech company ...
David Schatsky: Yes, for sure.
Tanya Ott: What advice then do you have for folks? Obviously, it'll be probably several years before this trend has a really significant impact on some industries or most industries, but what can company leaders do now to position themselves to leverage the technology?
David Schatsky: The first thing is that it's really important to be open to the reality of this trend. That tech, digital, and AI are not something that's happening to others. They're happening to you. And specifically, there's a set of implications that are worth considering. If you're a leader responsible for operations at your company, it really makes sense to start looking at smart machines, smart products, and how they can change the effectiveness of the operations of your company for the better, whether you're in logistics, manufacturing, or what have you. They can provide real significant boosts in efficiency and reliability.
Operations folks need to be aware of this and start looking at it. Product folks—product strategist and product marketing people—need to look at their product portfolio and think, what does the smart version of our product look like? How would a smart version developed by somebody else affect our market? And how do we deal with the opportunity and the threat here?
The strategy folks need to reflect on the broader economic dynamics that I mentioned before—you know, that the pool of money that's spent in an industry is going to shift around; it's going to change hands differently when smart products become mainstream.
And then the last thing I'll mention is that the real risk side everybody's talking about—the risks associated with technology, and especially with AI there's ethical risks; [people are afraid that] autonomous machines are going to run amok. So, risk leaders in organizations need to think through this. What can go wrong with intelligent products? With how they manage and use data or when they malfunction? Who's responsible and how [do you] mitigate the risk of all that?
Tanya Ott: There's also sometimes that harder nut to crack, which is the culture of an organization. There's a lot of fear built in with some of these AI products that they'll eliminate jobs, that there's going to be a shift in the kinds of work that humans do and that might leave some humans out of the process.
David Schatsky: Yeah, it's a really popular topic. People have been talking about it a lot over the years and I think that the only antidote to fear is better understanding. Whether you're a leader of an organization or you're a worker in an organization, it really will pay to become informed about how this technology is affecting how you do your job and how it can help you do your job better. For every story about workers being displaced by technology there are other stories about workers being empowered by technology. And I can’t comment on how those two sides balance out against each other, but [I will say] construction workers don’t work with picks and axes anymore, but there are still lots of construction workers and they use power tools. This technology can be thought of as a power tool to help people work more effectively.
Tanya Ott: David Schatsky is a managing director at Deloitte LLP and leads the firm's central trend-sensing efforts—a fancy way of saying they identify emerging business and technology trends that are going to impact us all in the years to come. His team’s report, Pervasive intelligence: Smart machines everywhere, is available at deloitte.com/insights. You’ll also find out podcast archive—lots of really interesting stuff in there including a longer conversation with Calvin Perry—the farmer we visited in southwest Georgia.
You can also find us on Twitter at @deloitteinsight (no S) and I’m on Twitter at
I am Tanya Ott. Thanks for listening!
This podcast is provided by Deloitte and is intended to provide general information only. This podcast is not intended to constitute advice or services of any kind. For additional information about Deloitte, go to deloitte.com/about.