May 2, 2017
by Maciej Kranz
In the first two installments of this series, we looked at some of the ways artificial intelligence (AI) and the Internet of Things (IoT) work together to create new business value. In this last article, let’s focus on some business and technology aspects of the AI and IoT partnership. Like twins separated at birth, they share some surprising similarities—but also have some important differences.
Do Similarities and Opposites Attract?
AI and IoT have quite different histories, but are in similar places in their evolution. Both began with a focus on improving legacy systems. For IoT, that meant improving and automating existing processes and infrastructure for better efficiency and productivity. For AI, the initial applications were focused on improving existing human-centric processes. Today, we are moving quickly toward an integrated IoT-native and AI-native approach—designed from the ground up for transformative digital methodologies. In many cases, new IoT solutions have AI capabilities built in from the beginning.
This convergence has spurred an unprecedented rush of new solutions, disrupting existing norms, but also creating opportunities for new work roles, new business models, and new value propositions. A few examples:
AI + IoT both enable automation, which is disrupting the labor market by creating a demand for new and different skill sets in many industries. In the U.S., the industries most transformed by these changing roles are accommodation and food services, manufacturing, agriculture, transportation and warehousing, retail, and mining.
AI + IoT are driving the move in many industries from product to service-oriented business models. IoT gives business leaders the data they can now use to make intelligent trade-offs between let’s say building a new plant vs. buying only the capacity they need. AI provides the intelligence needed to make those choices.
AI + IoT create new value propositions. Take the example of Simultaneous Localization and Mapping (SLAM)—I know it is a mouthful—in drones. SLAM allows drones to comprehend unknown surroundings literarily on the fly—even in dark, obstacle-filled environments beyond the reach of the Internet and even GPS. Using SLAM, drones can fly into dangerous situations, such as buildings damaged by fire or natural disaster, to check for people who are hurt or trapped. With real-time machine learning built into IoT devices, SLAM has become one of the most important drone applications in safety, security, and surveillance.
Why is this explosion of new AI and IoT developments happening now? Here is where some interesting differences emerge.
IoT has been around for a while in niche applications. However, it only came into the limelight a few years ago when three things happened:
Lines of businesses (LoBs) emerged as key technology buying centers, breaking technology solutions away from the exclusive domain of IT and shifting the focus to solving critical business problems.
Traditional markets such as manufacturing and transportation saw an accelerated transition of business structures away from closed, custom, vertically integrated, single-vendor approaches toward open systems with best-of-breed elements from multiple vendors forming cost-effective and modern solutions.
This need for multi-vendor interoperability and better cost structures then drove the adoption of open standards.
Unlike IoT, the emergence of AI has been driven more by technology revolution than business evolution. As we know, deep learning capabilities have accelerated in recent years, enabled in part by the availability of more and more powerful hardware. And in the process, AI has also become more relevant to business, powered by a flood of real-time data coming from IoT systems.
The Path Forward: Technology Implications
Two important technology trends set the stage for the accelerated evolution of AI + IoT.
First is an architectural shift from centralized to distributed clouds. Such “fog” systems extend cloud capabilities including data processing, real-time analytics, security, and network control all the way to the edge of the network, where the IoT-enabled end-devices reside and generate data. Since it is often impractical to send terabytes of raw data from let’s say a car or an oil rig to the cloud for processing, “fog” allows the data to be processed locally and only the results, exceptions or alerts are sent to the cloud—where centralized AI systems traditionally reside.
As use cases and architectures mature, I expect AI implementations to follow the IoT path and become more decentralized as well. If the logic is already set, AI-based systems such as predicative maintenance can be deployed in specialized fog nodes running on FPGAs or eventually even ASICs. Such fog implementations will dramatically reduce the costs and complexity of IoT + AI based solutions, accelerating their adoption and business impact.
The second major technology shift is the move away from proprietary systems to open standards. As we discussed, this shift has been going on in the IoT world for years. IT and operational technology vendors and customers have joined forces in horizontal and vertical standards bodies as well as key consortia to evolve legacy technologies and existing standards to meet the IoT requirements. Many of these efforts are critical for AI as well. For example, there are currently efforts underway by ODVA, Profinet, OPC, and other standard bodies to define data fields in sensors and actuators, and to standardize the meaning of the data they generate or consume. Having standardized ways to express “temperature,” “pressure,” or “rotational speed” is key to automating and driving down the complexity and cost of IoT + AI systems.
The Path Forward: Business Implications
While technology is important, managing change may be your biggest challenge as you consider implementing AI + IoT solutions. AI + IoT will have a far-reaching impact on organizational culture and the ways you hire and develop your workforce. You’ll need to look beyond the “usual suspects” for the skills you need, and you’ll need to think differently about developing your existing workforce for their future roles. Just think of the auto industry, which only a few years ago offered mostly blue-collar manufacturing jobs.
Today, every new vehicle rolling off the assembly line is a “datacenter on wheels” with millions of lines of software code running in it, and tomorrow it will become an AI-enabled fog node. Now, auto makers are chasing the same talent pool as tech companies. Inevitably, this will lead to a culture clash that can only increase as these companies move toward AI and driverless cars. Bringing together all of your employees in a coalition of the willing will be your first step in building a 21st century workforce.
If you are working in artificial intelligence, now is the time to bring IoT into your thinking. After all, most of the data you use in your AI systems is generated and delivered by IoT. So don’t take these data sources and underlying data distribution systems for granted—integrate them into your designs from the beginning. Start small and build on your successes. Pay special attention to security, change management, and building the right team of both internal stakeholders and external partners.
After decades of independent development, AI and IoT are together again at last. With the right technology foundation and a clear focus on the business side, they are perfect partners to help you digitize your business.
 Mizuho Securities, MI-Tech Vol 46: AI & Deep Learning; Primer to a Revolution, January 18, 2017.
Maciej Kranz is VP of Strategic Innovation at Cisco
About the Author(s)
You May Also Like