As Shakespeare might say if he was still around today, there is “much ado about nothing” when it comes to the Internet of Things (IoT). Companies are eager to talk about how Artificial Intelligence (AI) is transforming IoT by connecting devices, automating tasks and analyzing large chunks of data. But the truth is that while there is a lot of smoke and maybe some sizzle, there’s not a lot of real meat when it comes to significant plans for implementation.
There has been lots of discussion about connecting all manner of devices to the internet from your toaster at home to mission-critical systems controlling our power grid. However, Wi-Fi connected devices that enable us to remotely turn our lights on and off, adjust the thermostat and see what’s in the refrigerator are not true intelligence. Rather, they are time-saving devices akin to a TV remote control. Instead, devices need to get, share and act upon real data across a fully integrated network, otherwise they are like a remote control without a battery – useless.
And this is where the AI-IoT hype and implementation part ways. As soon as you start thinking about the data that has to cross a network from your “things” to an AI, you need to know what kind of data you are looking for – data that you need AI, rather than just a hard-wired business rule, to do something with. In other words, you need to know what problem you are trying to solve.
Addressing the connection between real business problems and technology solutions isn’t a new concept, but it may require nontraditional thinking. It was this kind of approach that led to the first electrical telephone switching system. In 1891, a funeral director named Almon Strowger was losing business because human telephone operators were sending calls to his competitors instead of him. His business problem was the direct result of an inefficient network architecture. Frustrated with the situation, he invented an automatic telephone switching system that allowed people to dial each other directly, eliminating the need for telephone switchboard operators. He patented his invention and eventually sold it to Bell Systems. His technology became known as the Strowger Switch and it altered the course of the telephone industry well into the twentieth century.
It is this kind of “out-of-the-box” forward thinking that will revolutionize AI and IoT. Strowger had a business problem and he built a new capability into the network to address it.
Injecting new intelligence into the network architecture
We are all familiar with the latest TV ads in which IBM’s highly-touted AI solution, Watson, “…reported that the AE-35 unit in the elevator may fail tomorrow.” However, this really is neither IoT nor AI. Rather, it is an automated machine response that relays a message about a potential pending failure. True intelligence involves more than simply reporting an outage—intelligence built directly into the network would tell you why something fails and even suggest new designs or other ideas to make it better. It might even seek other resources available on the network to heal a small problem before it grows.
In order to understand how AI and IoT work together, we first need to consider how both human and machine communications occur. In human communications there are four paradigms: one-to-one (telephone, email), one-to-many (blogs), many-to-one (Wikis) and many-to-many (websites, internal bulletin boards and forms). Similarly, there are four paradigms in machine communications: person-to-person (via technology), person-to-machine (Advise system status – are you empty?), Machine-to-Person (Alarm – alert-send map – I am full, come fix me) and Machine-to-Machine (Initiate upgrade – here’s the fix). AI can play different roles in each of these modes. Understanding where and how it creates value takes some thought and experimentation and the realization that it might not be effective everywhere.
Companies are investing in AI technology based on the promise that it will fully automate and improve communication throughout their networks. However, while AI’s “expert level” assistance has the potential to enhance communications in some areas, it still falls short in many instances.
Take, for example, a major power company that is responsible for maintaining many earthen dams. The company is concerned about determining if and when a dam would fail. Their resident “dam expert,” who is responsible for providing guidance on this issue, is about to retire so the company decided to look into employing an AI system to do the work. They hired AI “knowledge engineers” to follow and observe what the dam expert knows and gather expertise that they could build into an AI system. After many months, the knowledge engineers gave up because the dam expert’s experience, which often relies on his own “gut instincts,” does not easily convert into something that a machine could use. As it turns out, the complexity of human intuition is an elusive concept that is difficult to integrate into an AI system.
Scenarios like this suggest that we need a new approach to both AI and IoT. Without a defined problem, you can’t define a solution.
Rather than trying to focus on artificial intelligence, we would be better off referring to either automation intelligence or access intelligence. This puts the concept of AI and IoT to where they are needed most—embedded within the network itself. In terms of automation, devices need more intelligence to be of real value to the user and the provider and they need to be able to solve more than one problem. For instance, some type of AI-controlled “bots” in the network could automate maintenance and support. Security might be enhanced by similar “things” that monitor network traffic and police access to mission-critical resources to shut down intrusions before they propagate. These are problems that intelligent systems can be designed to address, rather than expecting AI to simply replace a human being with 30 years of experience.
Companies like CenturyLink understand the need to build intelligence into the network. CenturyLink’s extensive team of technology experts work with organizations to identify real-life problems and business processes that will benefit from integrating AI technology and IoT.
You can stop talking about AI and IoT, and starting doing AI and IoT.