Technology innovation has been the core mission of ABB for more than 130 years. As Chief Digital Officer of ABB, Guido Jouret, who oversees ABB Ability™, the company's unified, cross-platform digital offering, is chartered with envisioning ways to take ABB's pioneering digital solutions to the next level. In this interview, he discusses the opportunities ahead.
What changes has the world seen in recent years, and how do those changes relate to ABB’s current mission, and yours?
The biggest development has been the arrival of the Fourth Industrial Revolution, which merges the physical and digital worlds. It’s our job to build on the accomplishments of previous Industrial Revolutions, which over the last century improved life drastically, and not just in the First World. Life expectancy doubled, hundreds of millions of people rose from poverty, more people have medical care and access to food and clean water. But we’ve paid a price, and now our commitment with Industry 4.0 is to use all the digital tools at our command to expand the benefits of earlier Industrial Revolutions to even more people while reversing the negative consequences. That means greening the grid, electrifying transportation, fighting pollution, combating climate change with smarter infrastructure and helping solve the puzzle of feeding a planet of nine billion people when we have run out of arable land. I believe we can do it, and so do our customers. With ABB Ability TM, our unified, cross-industry, digital offering.
You've said that, instead of settling for incremental advances by optimizing existing architectures, businesses can achieve much more by changing the game entirely. What do you mean?
I recently gave a talk where I said that, as the Fourth Industrial and Energy Revolutions accelerate, customers are unlocking value by reinventing their business models in radically new ways via digital innovations. I use the phrase "Move Bits, Not Atoms" as shorthand to describe one way that the operating model for the industrial economy is being reenvisioned.
Today, we move "atoms"—vital resources such as energy, water, food—over incredible distances, to deliver them from their sources to where they're needed. But that's come at the cost of heavy environmental impact, via increased carbon emissions and deforestation. What if, instead of transporting "atoms," we could deliver expertise, or "bits"—such as know-how encoded in software—in a friction-free manner? Bits of information are essentially weightless—a water molecule is heavier than an electron by a number that’s got 30 zeroes after it. So the idea is to move those "bits" to where the resource "atoms" are needed.
This isn't just an obscure analogy. We see it happening already in energy, where we're moving from large centralized power plants to distributed generation using solar panels, microgrids and wind turbines. So, small is beautiful!
It sounds like you're talking about the rise of distributed intelligence.
The model of the future is going to be more modular and decentralized. However, moving away from today's larger, centralized model requires coordination, because instead of having one big power plant or factory, you will have lots of smaller ones. We'll need to get these smaller units to play together. It's a bit like conducting an orchestra. For example, with multiple sources of energy generation—from turbines, solar panels and battery storage—we will need to have orchestration between supply and demand. What I'm essentially saying is that we need a digital nervous system.
What is a digital nervous system?
It's not going to be possible to centrally manage and control everything. You can't have all the information arriving in a big control room somewhere, and have a person making all the decisions. We'll have to rely more on each individual system's "autopilot." That is, we're going to have to offload day-to-day, micro-level decision-making to increasingly intelligent systems. So, in this digital nervous system, there isn't a single pilot in the plane. There are many pilots—they're decentralized, but they're coordinated. They're also designed to failover gracefully. A great example of the latter is the way the internet works.
What role will artificial intelligence play?
I have an optimistic view of AI. For one, we've made tremendous progress in machine learning technology over the past several years. But what most people don't realize is that there has been no fundamental improvement in the underlying neural network algorithms. The models we use today are the same ones from 40 years ago. Because we now have so much more computing power, we can build bigger neural nets. But we're still replicating a very primitive approximation of a human neuron.
Just to give you a sense of the scale of the challenge, consider the computer that plays GO. It's actually playing the game using a very large server farm, which consumes thousands of watts of power. The average human brain consumes only 60 watts, which is equivalent to a traditional incandescent light bulb. And the brain isn't just playing GO, it's also processing visual information and managing your body.
Another reason this disparity in energy usage is notable is because human neurons are thousands of times slower than electronics. In theory, a computer should be a thousand time more efficient at running the GO algorithms than our brain is, but it's actually millions of times less efficient.
So does that mean we haven't got the architecture right?
Exactly. We're trying to approximate something using crude instruments. Imagine what we could do if we understood how human neurons really worked! That's why I think AI has a fantastic opportunity ahead. But the rate of progress that we can make in AI is potentially millions of times what it is today.
Robots seem to be advancing very rapidly, what is your view?
Robots are going to come in all shapes and sizes—everything from floor-sweeping Roombas to automatic guided vehicles (AGVs), which are essentially robots on wheels that work side by side with people in, for example, warehouse settings where the task is to get inventory off the shelves and put items into bins or on conveyors for shipment. We'll also start to see robots in the service industries—robots will cook, cut hair, dispense medicine.
I also see a bright future for robots and automation machinery in agriculture. Consider that, in Japan, the average age of a farmer in 2017 was 67 years old, so there's obviously a pressing need.
I think we have to recognize that robots will not look anthropomorphic. They will not necessarily have a human shape. Consider that an autonomous ship, which we will see in operation within the next year or so, is a form of robot, because it's got intelligence along with motors, drives and actuators.
What are the challenges ahead?
Even with all the progress we're making in AI and machine learning, there's still a fundamental limitation, which is that we're creating systems which are inherently brittle. This means they can only solve problems that are contained within the training set, or set of examples and use cases, on which they've been trained.
A good example is the driverless car. If you instruct a driverless car to never cross a double yellow line, that works fine until the car gets stuck behind a truck that's not moving because the driver has stepped out to make a delivery. That autonomous car will stay happily stuck behind that truck, and traffic will get backed up.
So we have problems with exceptions or corner cases, as with this example. How do you apply judgment here? Because the right answer is for that car to inch across the double yellow line so that it can "look" past the truck, and if it doesn't see oncoming traffic, it can then pass the truck. This raises the whole issue of judgment and forecasting, but it's beyond the scope of current systems. They're not able to do that.
You've spoken frequently of the ascendance of operational technology (OT) and what that's going to mean for our industry.
The industrial space, where OT lives, is becoming a massive beneficiary of consumer-led innovation. Let me explain. Years ago, if you wanted to work with the best technology, particularly in the United States, you went to work for something associated with government or NASA. Next, with the advent of microcomputers, the pendulum swung towards the enterprise. Then came the Internet and people started to get connected. But the home experience was based on dial-up modems, and it wasn't good, whereas at work it was better because you had always-on internet access.
Since around the turn of the 21st century, the pendulum of innovation has shifted to the consumer side. Today, if you want the best technology, you get it at home. The phone in your pocket is much better than the one on your desk at work. Your home PC is probably much more powerful than the your work-issued laptop. Virtual reality, video games and mobile phones are certainly examples of innovations propelled by the availability of large potential consumer markets.
I'd like to note that "consumer-led" does not mean there's no business value—quite the contrary. Take drones—they're a massive multiplier of employee productivity. We can put methane sensors on drones and fly them around refineries to do faster and more thorough inspections. We can inspect oil and gas pipelines for leaks. We can fly drones up to take pictures of wind turbines.
You once wrote a paper discussing techniques for coming up with the next big idea. Tell us about that.
My favorite way of looking for the next big thing is to find evidence of tinkering. Let me explain: tinkering is when somebody is trying to solve a problem, but the technology they're deploying is obviously inappropriate for the task. It's the equivalent of using duct tape and baling wire, and your first instinct would be to assume it will never work. The point is that the problem you're attempting to solve must be very significant, if you're willing to put up with a kludgy solution. So evidence of tinkering—where someone has devoted actual time and effort—is a much more effective technique in identifying areas ripe for research than is the typical approach of interviewing people.
What do you see as the major issues going forward?
The big four topics are energy, transportation, water and food. That's because they represent collectively the bulk of our planet’s operating system. We've talked about some of the innovations in energy—solar panels, microgrid—and the development of autonomous vehicles in transportation. As the world's population continues to grow, water and food are going to be major issues.
Around 98 percent of the water on the planet is actually saltwater, so desalination needs to be part of the solution. As for food, I'm passionate about the possibilities for urban farming. It’s really clear that we have to reinvent agriculture as we know it. There isn’t enough land, there are going to be many more mouths to feed, and most of the people are going to be living in cities.
When you think about it, urban farming aligns with what we talked about at the beginning of this interview—moving to a distributed model. Rather than shipping food vast distances, we can work more effectively together on a local basis, because the cost of coordinating and communicating has essentially dropped to zero, thanks to enabling digital technologies.
