4th Industrial Revolution: a Human Control Imperative

As has been pointed out time and time again, we’re in the midst of a fourth “Industrial Revolution,” often called the “Digital Revolution.” What is it, though? Why did it arise, and what does it mean? Klaus Schwab in his GE Reports article originally in Foreign Affairs on the Fourth Industrial Revolution, described the first three industrial revolutions in a perfect framework:

• The first used steam and water to mechanize production
• The second used electric power to create mass production
• The third used electronics and information technology to automate production

However, he neglected to use the same framework when describing the fourth revolution. He did so because to put the fourth industrial revolution in the same context, it would have to be described as “harnessing the power of the digital revolution to evolve our relationship with production.” It’s a scary thought to admit that we as humans have been behind the curve of our own progress, but it’s the truth.

To put the three previous revolutions in context, let’s consider the process of weaving. Weaving began as a manual process of interlacing threads into a pattern that was durable enough to stay in one continuous piece, and function as something other than its constituent parts. It took time to complete a single woven piece of fabric, but the process was entirely under the control of the weaver.

Then the loom was invented. It mechanized the production of woven cloth. Deceptively simple, it streamlined the manual activities of weaving into a mechanical framework that allowed the weaver to avoid mistakes and keep things straight throughout the process. The important thing to note here is that the loom distanced the product from the maker. This imperceptible shift from intimate relationship to one where a human guides as opposed to makes is the first crack that created the chasm we’re dealing with today. But come on, we’re talking about a loom here, how big of a problem could it cause to automate? The loom had evolved such that a person could direct the making of a woven cloth with more organization, precision, and ease, but we’re still looking at a 1:1:1 relationship — one weaver, one loom, one cloth. That ratio persists, no matter how many looms you put in a room or factory.

Enter the second industrial revolution, where suddenly electric power can add cogs, gears, and a whole lot of “oomph” to the loom. The automatic loom magically breaks free of the 1:1:1 relationship, making it advantageous to cluster as many looms together in one big room to run off of a single power source and begin making woven cloth en-masse. Now we’re looking at a system we can all recognize — humans managing a process beyond their manual control. After the Second Industrial Revolution one human can operate 10, 20, maybe even 100 looms at a time. 1:100:100 is a whole different ball game, folks.

When the ratio is 1:100:100, that one person needs to operate under extremely controlled conditions to avoid breaking down. Similar to a top fuel dragster in a drag race, there is very little adjustment done on the fly — too much is happening. Ninety-nine percent of a drag race goes into the planning stages to prep the vehicle, control for variables, set boundaries, and limit the driver’s need to do anything drastic. This dragster is automation. Today, set a software program in motion, and it’s as hard to course correct mid-run as a missile-shaped car doing over 330 miles per hour. So, the Information Technology revolution is what builds us the weaving factory capable of performing like a top-fuel dragster, and with the same basic principle in place — plan everything down to the last tiny detail so that when you flip the switch — as few things can go wrong as possible.

The important thing to note here is why we’ve worked so hard to reduce variables and streamline every bit of the process. It’s because, at some level, we’ve recognized our inability to control the process. Humans have been evolving our ability to produce more without evolving our ability to control alongside it. Consider the weaving factory of the third revolution. One human could theoretically operate 100+ automatic looms at maximum pace for as long as raw material is available. I say theoretically, because we all know there needs to be a number of workers on the floor to maintain the machines, find problems when they occur, or even just manage the reloading of raw material. If that’s the case, we’re going to need even more workers to ensure that the streamlined, automated, mass-produced output doesn’t have a flaw. This quality assurance (QA) process is required at the end because, given the impossibility of course-corrections, errors are the logical result of a process that can run at insane speeds all by itself. Even a single flaw in the process would multiply and reproduce as effectively as the rest of the process, rendering everything flawed.

And that’s where we are today, on the precipice of this fourth revolution. The revolution that brings us humans closer to the things we’re making.

For illustrative purposes, let’s take that weaving factory with it’s 100 looms and one manager, then contrast it to the original weaver. The weaver was painfully slow and error-prone by comparison. However, when the weaver made a mistake in a stitch, what happened? A few moments later he’d realize it, backtrack, and properly re-do the weave. By comparison, what happens when a factory full of automated looms gets a weave wrong? Dozens of humans panic and raise alarms in QA, the factory stops, an assessment is done, and the ruined inventory — likely hundreds or thousands of woven cloths — is surveyed. A team investigates what caused the error, it’s fixed, tested, and anywhere from a few hours to a few days later, the whole thing is back up to speed. Today, we’ve created processing power beyond our meager ability to control.

This is where the fourth revolution comes in. It brings us closer to the process itself through the use of connected devices, self-monitoring sensors, feedback loops, and independently operated units. We are finally able to evolve our interaction with a process such that we can control it while it runs. No, our ratio won’t be 1:100:100, but it might be that four or five people can each track the intimate workings of some such portion of the process, or a smaller subsection of independent machines. With lightning fast, intimate, detailed information, an operator can make changes on the fly, halt a process efficiently, and still restart quickly.

The days of “set it and forget it” are gone, but I for one am not sad. It’s scary to admit how far we’ve come and how many of our impactful automated processes not effectively managed by our top-down control. Taking the next step opens us to a world of possibilities.

The Ted Radio Hour on “Getting Organized” perfectly describes the conflict we have today. In the most chaotic natural systems, there is an incredible amount of order and organization, but in all cases it is not centrally controlled. Humans have pushed the limits of central control as far as possible within business, governments, and especially industry. Where we need to hire other humans is only at the lowest level, because the task is too visual or too spatial to design an automated system to perform it at scale. This connects workers to mere pieces of a process, without taking ownership of an outcome. As “Getting Organized” points out, most naturally occurring complex systems are actually self-directed. We find the same sort of decentralization happening in software today. We call it agile, but it’s really a decentralization of the control with a re-centralization of focus.

What connected devices like the Internet of things, big data, cloud infrastructure, and impressive advances at the micro and macro level are letting us do is trade little bits of control for truckloads of clarity. By giving up on the idea that a process can be perfect and faultless, we accept that consistent course corrections must be made. By advancing the way we collect, connect, and utilize intimate knowledge of each piece of a process, we allow for more flexibility to pause, adjust, or completely redirect our energies. We bring back the middle class of workers who are independent within a moderate scope, who add value, feel purpose, and can learn and grow in their careers. We naturally then look to artificial intelligence (AI) to replace humans at the lower levels of work. Driving someone to the grocery store has never been a prestigious job. Neither has mechanical maintenance, appointment setting, or general admin work. These are all the initial domains of AI.

Get to the point right? Ok.

The fourth industrial revolution is one where we humans admit we over-reached on automation without intimate control. We’re going to look back at every “perfect machine” we’ve built over the last 200 years and decide it’s actually a “terrific machine” that we need to better understand during operation. In order to do that, we’ll need to employ ever-more sophisticated and integrated sensors, monitors, connected devices, and microcontrollers to bring us intimacy with a machine or process at scale. Each of these new devices and sources of intimate information will further accelerate our collection of “big data,” which in turn will serve as fodder for nascent AI to become the factory worker of the future. Far from being a human-dominating force for evil, AIs will be diligent, focused, efficient workers, who understand specific processes intimately and can react at the speed of electrical impulses to correct errors and control processes.

Imagine that a jet engine mechanic had the power to actually see everything that was happening inside an operating engine. From fuel flow, temperature, and output, to friction levels, lubricant viscosity, and individual component stress or vibration. Wouldn’t you feel infinitely safer if you had one mechanic with this kind of superpower, focused on and dedicated to each of the engines on your next flight? I sure would. Would you fear for the fate of humanity at the hands of these specialized mechanics? Most certainly not.