The Machine Intelligence Revolution Takes Shape
Three disruptive computing technologies dominated the thinking of technologists in 2020:
Quantum computing research is still a slow, complex journey into a magical world. But we've seen breakthroughs this year. Admittedly, the technology is still in what we call “the vacuum-tube era,” but several companies are demonstrating very rudimentary “quantum supremacy.” That means quantum computers are finally beginning to do what no conventional computer could ever hope to do.
Similarly, 30-plus new semiconductor technologies used to accelerate the computing of various workloads with much less energy are emerging from research labs. These architectures are especially important if we are to cost-effectively realize the promise of AI & machine learning. Their impact on price-performance is expected to unleash new commercial applications that were nowhere near cost-effective just a year ago.
And in the 5G world, one of the few happy surprises that occurred during the COVID-19 crisis was people's acknowledgment of four things:
Importantly, 5G will be built differently from legacy telecom. It will use open hardware, software virtualization, and containerization. And it will be a heavy consumer of AI and machine-learning technology. So, telecom development will come to focus on the same issues as the rest of the U.S. technology industry. That is, the actual technology that is used to build 5G and beyond is going to be much more dominated by widely used IT and cloud technologies than "legacy telecom" has been.
Notably, industry and technology are just now getting to the stage to build many game-changing solutions that visionaries imagined 30 years ago at the beginning of the Dot-Com era. For instance, we don't yet have automated delivery drones flying over our cities intelligently knowing how to bring us our goods and services without killing anybody. We don't have self-driving cars. We don't necessarily have smart cities. And we don't have really smart factories yet. But we have substantial work-in-progress toward all of these.
Those and hundreds of other transformative solutions are still in front of us. But we haven’t been sitting still. We have enough evidence from the early waves of “smartifying” the world, to know that the big problem is not making the device smart, it's making the device smart and efficient within a scalable system.
As spelled out in Ride the Wave, if the solution was formulated as a standalone, fully self-sufficient, hyper-intelligent entity, you wouldn't have enough resources to make it do whatever it's supposed to be doing. To address this problem the so-called “edge computing layer” has materialized, not so much as just an interesting place to do IT, but as an offload mechanism for the “smartification” of the world.
Consider the case of a cell phone, Augmented Reality goggles, or other mobile devices that, instead of doing all the video processing on the device, push about 80% of that processing into an “edge compute layer” that has all the power it could need; the result is that now you have a highly efficient Augmented Reality experience on a mobile device that's getting the assistance from the edge. But more importantly, it actually far exceeds its original capability because it's effectively tapping into infinite computing power in the cloud. So, it has more artifacts, better video resolution, and greater color depth.
What is this telling us? The mobile device by itself isn't the answer and the cloud by itself isn't the answer either. The answer is this combination of “cloud-based infrastructure” plus “edge infrastructure” plus “the front-end devices,” all working together to deliver an unprecedented balance between cost, functionality, feature set, and deployment.
But it goes further. The network of devices and their related infrastructures can create even greater value by operating to complement each other and the whole. Beyond technologies becoming better, smaller, and faster, this also means that at the edge, the device that you have in hand is part of a wider “network mesh.” So, AI can extend out from the cloud so your device and the other devices in the network can be made smarter.
Consider what we’re learning from all of the work done in autonomous vehicle development around the world. Obviously, the car itself is going to be quite smart. And each car can sense the cars around it as well as the road surface. So, imagine if all those cars started to not just share their long-term data, but their immediate view of the world around them in real-time; that includes sharing it to nodes that were adjacent to them in real-time so that your “road network” itself had a master real-time understanding of all the cars close-by on the road. As a result, your car could tap into this intelligent road via the edge compute layer to let it see around corners. It could see everything you can't see, as well as what other people can see. That means your head-up display could provide a visualization of everything around you because of that collaborative compute model. That's an incredibly powerful tool that isn't possible if the device is trying to solve this problem by itself.
Importantly, this applies far beyond autonomous vehicles. In fact, you can transpose that into many other industries. But the autonomous-driving one is particularly fascinating because you will have a very smart and robust device that can operate all by itself, but it can operate far better in many dimensions when it can tap into the collective consciousness of all of the cars, and all of the roads and all of the things around it in real-time. And the only way to get this real-time responsiveness is by tapping into an edge computing layer.
Such a structure is going to become one of the big breakthroughs because it delivers a collective understanding in real-time which is local to you.
This brings us to an important distinction that we’ll see across many industries and business sectors as the Golden Age of the Fifth Techno-Economic Revolution unfolds. When we apply machine intelligence to anything, whether it be a self-driving car, a business process, a user experience, or whatever, there are two types of success.
One type of success comes when we completely revolutionize a business and turn it into something that has never before been delivered; that would include a level-five self-driving car or a fully autonomous household robot. That is a big, big jump, and it's worth taking that jump—it just takes a very long time to get there.
The other form of success in terms of harnessing machine intelligence is when a business augments the cognitive tasks that human beings typically do. In the pre-AI world, “you were on your own.” It was up to each person to make that decision. Very rarely did they get much help on the thinking side. They might get a lot of data, but they had to sort through it. The recommendations did not really come from technology; you had to figure it out.
What’s different in the new world? By careful application of machine intelligence to places where human beings have to take data, understand it, and make a decision, we can accelerate the process or make it less prone to error.
What companies and researchers are finding as they examine business activities ranging from supply chain processes to predictive maintenance processes, to radiology systems, is that a multitude of 5% and 10% AI-based improvements in getting aspects of the process to work better really adds up over time. And every time these improvements collectively improve a supply chain by 5% or 10%, or radiological accuracy by 20% or 30% in detecting tumors, that's a very powerful outcome, not just to an individual, but potentially for society. And that’s especially true because the cost of making each improvement is typically small.
Obviously, big breakthroughs are great. But there's so much more than business can do with this technology as we seek to enhance every process where human beings have to make decisions, and augment those with machine intelligence to make those decisions more accurate, more speedy, and more likely to have a positive outcome.
And that’s especially encouraging because capturing so much of this windfall for consumers and shareholders doesn't require massive breakthroughs. It leverages the technology we already have today. And every time we do it, the process gets better, the cost structure gets better and the outcomes get better.
Given this trend, we offer the following forecasts for your consideration.
First, over the next five years, intelligence will be embedded in almost every industrial process and most consumer products.
There really isn't an industry that is not contemplating this transformation. In some industries like health care, it's hard to implement AI quickly because it's such a regulated industry; as a result, the timeframes are very long. That’s why we’re seeing the early impact emerging in areas like drug discovery and the so-called wellness industry. For instance, there is the Oura Ring, which monitors your temperature and a bunch of other vital signs. It's a wellness tool that uses advanced machine intelligence to give you a pretty good early warning that you might be coming down with something before you know you're sick.
Second, machine intelligence price-performance will accelerate sharply as 5G networks roll out.
The hallmark of future telecom infrastructures will be automation. Intelligent machines will make the decisions around spectral efficiency, bandwidth tuning, and all kinds of other things because there's just no way a human being can run a network of one hundred-million subscribers; and in the U.S. alone, some of these mobile networks, 10 years from now, might have a trillion “things” connected to them. Most importantly, chips and algorithms developed for these infrastructure roles will provide a foundation for innovative applications in other areas.
Third, logistical applications will be another area dramatically transformed by machine intelligence in the 2020s.
There are machine intelligence initiatives going on in the freight and logistics space where people are realizing that goods and services move too slowly and clunkily. This begs the question, “what if we fuse intelligent forklifts, with visual surveillance, object mapping, and algorithms to decide how to load planes or trucks optimally?” Then, we can integrate logistic infrastructure that uses AI to identify bottlenecks in the supply chain. And then, let AI figure out the pattern and develop a set of logic around it. The payoffs could be enormous. And that’s particularly important if we’re trying to raise living standards and total wealth in a country that is rapidly aging. And,
Fourth, the biggest challenge to the machine intelligence revolution will be security.
Throughout history, every mechanism that has created wealth and power has attracted criminals and enemies seeking to steal or destroy it. That’s not going to change. And it’s the very technologies that create this wealth that are going to enable attacks and defenses. Quantum computing and Quantum networking will play particularly important roles in this battle. Furthermore, the ability to identify intrusions and meaningfully retaliate against perpetrators is vitally important. Therefore, there is no time like the present to seal our Technological Iron Curtain.