The Sentient Machine
1. THE EMERGING INTERNET OF THINGS
Today, at this very moment, a kind of membrane is growing around all of us. We can liken this to a planetary skin or even a cortex at the center of our entire built environment. This network, the Internet of Things (IoT), is growing denser and denser as ANI makes more and more of our world “smarter.” The many billions of man-made objects that we interact with daily—cars, stoplights, toothbrushes, bridges—are being transformed from mere static forms to objects with cognition.
Before we look at the immediate implications of this growing intelligence, it’s worth considering how other intelligence “explosions” changed our ancient and preindustrial civilizations. When Australopithecus, or ancient man, started building tools 2.5 million years ago, for example, the crude stone objects he formed served only to more effectively harness the power of human muscle. These tools were not imbued with any form of locomotion independent from man. Fast forward in the story of human evolution to approximately fifteen thousand years ago, with the domestication of cattle, and five thousand years ago, with the domestication of horses, however, and we see “man” seeking out more sophisticated ways to leverage or
augment his own muscle. This drive leads to the invention of mechanical devices such as the wheel in 3500 BC and the pulley in 1500 BC. In these innovations, the muscle-power of man is not just harnessed, it is magnified.
Although it took more than two million years, we finally graduated from crude tools shaped from stone and wood into developing systems imbued with their own source of power, independent from us. As long as the animals powering these devices were fed and the mechanisms maintained, we were able to use these systems to perform critical functions, such as raising water from a well and lifting large stones and logs. In the year 1698, with the evolution of the steam engine, we ultimately crossed a bridge into the beginnings of the industrial age. The steam engine enabled a means of making locomotion independent from any form of biology or nature. And in creating this locomotion, we could build bigger, faster, more resilient and powerful muscles than had ever been observed in nature.
Yet despite all this mechanistic prowess, these tools and systems always remained dependent on decisions made by us. We prescribed specific ranges of motion for them, and when one of them needed to be turned on or off, it was inevitably us, humans, pulling the switch. In the early nineteenth century, however, in a subtle shift of machine innovation, all that began to change.
• • •
It was 1801 and French weaver and
inventor Joseph Marie Jacquard was looking for a way to create more sophisticated textile designs. Up until that moment, any design beyond basic lines needed to be hand-stitched, meticulously constructed by artisanal craftspeople. Jacquard realized, however, that he could
bring a new flexibility to his sewing machines. He decided to teach them to interpret instructions, not just act out the prescribed sequence of movements that their mechanical design dictated. This idea revolutionized not just the textile industry, where it was used to weave a multitude of patterns on the same machine, but industry in general. The punched cards that encoded Jacquard’s designs—programs that defined patterns—were very similar to those used in computers a century and a half later. The act of separating form from function, or instructions from implementation, gave rise to the notion of programmability.
It wasn’t much more than a century and a half later that an entirely new discipline—computer science—emerged. Like the encoding in Jacquard’s looms, computer science evolved frameworks and processes for the efficient specification and execution of complex activities. This science concerned itself with ever-smarter ways of programming machines, and one of its subdisciplines—artificial intelligence—aimed to produce thinking machines entirely independent from humans, physically and mentally.
Today, 216 years after the Jacquard loom was invented, programmable computers the size of a fingernail can control powerful, miniaturized motors and obtain information from a plethora of digital sensors to sample, process, and respond to the real world. An ever-growing sophistication and intelligence in the programs that control these devices, ubiquitous connectivity between them, and a growing capability in the processors, sensors, and actuators to which they are connected promise to lead us into a future we can barely even imagine now.
Welcome to the Internet of Things revolution, an era when intelligence will be embedded everywhere, when synthetic
devices and systems will make a growing number of decisions on their own. In this age of IoT, there will be billions of devices communicating with each other: negotiating, interacting, measuring, responding, and initiating all without any human input. In an effort to explain how I see this future evolving, I will paint a picture of IoT adoption in three waves.
THE FIRST WAVE OF IOT: MEASURING AND TRACKING
We are already firmly in the midst of the first wave of IoT. On the consumer side, we have wearables and gadgets that measure our pulse and heart rate, track how much we’ve walked over the past day, attempt to guess our circadian rhythm and activate an alarm when we’re sleeping lightly, and that automatically send pictures of our home to us when they suspect someone is at the front door—or someone is trying to break in.
On the business side, we have sensors embedded in almost every major industrial asset—from generators and turbines, to pumps, grids, and drilling equipment. These sensors are being used to gauge the more obvious aspects of a system’s performance. They measure things like temperature and pressure and store these measurements for subsequent human analysis.
THE SECOND WAVE OF IOT: MODELING AND PREDICTING
In some areas, we are on the cusp of entering the second wave of IoT where data captured from first-wave devices will be used
by the devices themselves to model the environment, their own behavior, and the behavior of other systems to predict the future. For example, consumer wearables that simply monitor heart rate and pulse will evolve into wearable doctors that won’t stop at measurements, but will, instead, provide a full diagnosis as well as recommendations. In order to make this happen, a greater level of intelligence will need to be embedded in these devices, as will a larger number of sensors and environmental inputs. The cognitive capabilities of the devices themselves, or the networks they connect with, may include the ability to read and process natural language and inputs like photographs and video streams. Imagine a wearable that watches what you eat, figures out what it is, calculates the size and hence caloric intake, and uses that information to warn you of everything from relatively benign diet violations to the accidental ingestion of a food item that could trigger a life-threatening allergic response.
In the world of business, we’ll not only see machines monitoring basic elements of performance, but machines that will use these first-order data streams to evolve deep predictive models that look for higher-order interactions of measured quantities such as vibration, temperature, and pressure to uncover the complex physics that drive systems in the chaotic real world.
We’ll also see network-connected systems that don’t just sense but act in an increasingly sophisticated way. These systems will include delivery drones, self-driving trucks and tractors, and increasingly sophisticated factory and warehouse bots that use vision to detect objects and sort products and packages.
THE THIRD WAVE OF IOT: A TRILLION FULLY AUTONOMOUS DEVICES
In the third wave, the true potential of the IoT will materialize. We will have unlimited, easy to replicate, massively distributed, and federated network intelligence powering cognitive, fully autonomous devices. Sensors will become incredibly powerful not just because of the capabilities of the hardware, but because of the highly intelligent AI algorithms that will be able to fuse information from basic sensors into a coherent, granular, and complex picture of reality. This will offer a type of picture that goes far beyond what humans are able to build with their eyes, ears, smell, and touch. This will be a world that is perceived most profoundly by the intelligent devices that inhabit it. The humans who built those devices will be left, largely, unable to experience this reality.
This third wave of IoT will include autonomous and mobile systems that sense and avoid conflict in messy, real-world scenarios. Consider, for example, algorithms that empower fleets of hundreds of thousands of autonomous drones to carry out an ever-increasing range of functions for their human owners, from crop dusting to the delivery of emergency medical supplies to policing towns and cities to enabling the next generation of weapon systems in the form of autonomous hunter-killer swarms. As all of these activities power more and more of our built environment, the human starts to leave the loop. As we will see in the following chapters, this will cede decisions in our world to the burgeoning network all around us.