Automation is ancient technology for building things and putting them in motion.
Goal-Oriented, Code-Driven Work
Automation is an ancient technology for building things and putting them in motion. We think of it narrowly as a force for replacing human labor, but to prepare ourselves for what lies ahead, we need to think bigger.
At a basic level, automation is “goal-oriented, code-driven work.” Biology is thus the original automation. Its DNA coding embeds survival goals into carbon-based life forms. The plants that spring from seed each spring build themselves with biology’s automation, a force we describe as growth. The bird outside your window moves freely in physical space with the kinetic automation that we call animation or behavior. Life grows, heals, and reproduces without any help from us. This is the deeper meaning of automation that is found in the buzzes, groans, creaks, and whirs of nature.
Seeing automation’s roots in this way, prepares us for the next generation of goal-oriented, code-driven work. Machines are evolving into something approaching the uncanny autonomy of biology. We’re not there yet, but make no mistake, many of us will live to witness the rise of an automated intelligence well beyond our current understanding of automation.
Automation + Artificial Intelligence
Automation is inherently intertwined with intelligence. Biological automation relies on genetic intelligence, just as machine automation relies on machine learning. The simplest way to see this relationship between automation and intelligence is an “intelligent agent.” The agent is made up of three sub-systems: a sensor for taking in data, a model to abstract and make sense of that data, and an actuator that sets the whole system in automated motion. The sensor feeds data from the outside world into the machine learning model, which then increases the intelligence that then drives the system’s automation.
This integrated system for sensing, learning, and taking action largely mimics far more ancient biological systems. But these newer, technological systems scale in ways that make them qualitatively different. As the quantity of data flow through the sensor increases, the fidelity of the machine learning model increases. As the model improves, the effectiveness and efficiency of the automation grows. That increases the performance and attractiveness of the system to end users, increasing usage and data, which leads to yet better models and automation. The result is a virtuous cycle that creates an upward spiral in intelligence and performance—a fusion of automation and machine learning.
This feedback loop between automation and machine learning will create wonders so responsive to their surroundings that it will feel as though they are alive. But they will not be alive; just very intelligent.
This is important for us to remember.
Humanity’s Role in Automation
Despite the cosmic scale of these changes, humans still play a vital role in setting automation in motion. Our designs and engineering give rise to the technologies and processes of automation, and even after these contributions fall to automation, our goals will still guide its ongoing operations. We are the volition of these systems. One of the central questions before us is whether human governance will maintain this oversight function in our future.
The other essential function that humanity plays is that our subjective experience gives meaning to automation. The first generation of Earth’s automation exists in biological ecosystems, the second in economic systems. In market-based economies, survival is determined by economic “utility” that is ultimately determined by end-users. Even in command economies where economic decisions rest in just a few hands, human judgment still determines automated ‘fitness.’
This is a critical and often underappreciated point, and the deeper meaning of “end user demand.” Barring the emergence of some form of artificial machine subjectivity, humans serve as the subjective source of valuation for automated systems. We are not just their volition, we are what gives their growth and operation meaning.
The Automation of the Economy
Herein lies the simultaneous promise and peril of automation. For humans are deeply encoded with our own ancient survival coding that sometimes makes us operate in selfish and unsustainable ways.
Automation leaves a distinct footprint within each economic sector, from resources extraction to manufacturing, services and government spending. Its greatest impact on Earth’s ecosystems happens with resource extraction industries. As agriculture, fishing, and forestry automate, they accelerate extraction to increasingly unsustainable levels, risking widespread ecological collapse. Resource extraction systems feed the manufacturing, services, and government sectors, where poorly designed incentives and production processes lead to accelerating waste as plastic islands in the Pacific and a blanket of greenhouse gasses enveloping our atmosphere.
Just as automated resource extraction most impacts the planet and automated building most impacts technologies, automated services most impact human beings. Service industries largely depend on engagement with end users to generate their value, which means that service automation means the kind of automated self-service that we see with automated service platforms like Google, Facebook, and Amazon. These platforms operate at “Internet scale,” engaging billions of people around the world in their work. These systems have rapidly become humanity’s primary interface with machines thanks to the rapid advances made possible by their couple of automation and machine learning.
Governments play a variety of roles in automation. They can deploy it to improve flows of traffic on city streets and flows of energy into our homes, or use it to surveil us and even as the lethal force of automated drones. The global rise of authoritarianism is particularly worrisome given the concurrent explosion in automation’s scope and capacities right now.
Governing Machines
The unprecedented power, scale, and scope of automation on our economy will deeply shape our insides and our outsides in the decades and centuries to come. There is no guarantee that this impact will necessarily be benign. Our only hope lies in building resilient governance systems that ensure human volition and meaning making remain in the driver’s seat of these increasingly intelligent systems. This will take hard work, and there is no guarantee of success. But governing machines will be the last form of human work. It’s now time for more of us begin to turn our attention here.
Gideon, another very thought-provoking article, and very timely to a book I am reading which reminds me of so much of which you write. The book is “Lifespan – Why We Age and Why We Don’t Have To,” by David Sinclair, PhD. I am very early on in the book, but just this morning I thought of you as I read the comment “But the science is moving fast, faster now than ever before, thanks to robots that analyze tens of thousands of genes a day, and computing power that processes trillions of bytes of data at speeds that were unimaginable just a decade ago. Theories on aging, which were slowly chipped away for decades, are now more easily testable and refutable.” Your references to biology as the original version of automation ties in directly to this book, as Dr. Sinclair describes how our genes and DNA work, and the research into why we age. My wife is such a “researcher” of everything, and this morning I commented “What would you do without Google?” to which she replied, “I’d use the encyclopedias that I bought with my own money from a door to door salesman.” Wow, how our world is changing!
Thanks, Bill. Sounds like an interesting book. That topic seems to be on many people’s minds in Silicon Valley these days. And yeah, it’s interesting to think about proteins as kind of the automation that builds what it’s told by the DNA.
The comment about encyclopedias reminds me that my first job out of business school was working at Microsoft on our first multimedia encyclopedia. We shipped it on CD-ROM and were constantly going on about how it blew away print encyclopedias. Some of us there pushed to make the jump to the web on other consumer offerings (I built a car buying service of all things), but there was just so much invested in all that work that when Wikipedia came along it completely blindsided us. Yep, the world does not stand still. That’s for sure.
I just switched the Vital Edge website to Disqus for commenting. In the process, the following comment by Sowmyan Tirumurti was deleted:
Automation is looking for productivity and efficiency for the well to do. I guess there is less money to be made helping the downtrodden. But as a problem building up for the future, the ‘have not-s’ will have to be addressed too. Poverty is the lack of adequate marketable skills. Poverty needs low cost solution. Current automation answers low cost by eliminating labor. That is increasing the disparity. Also it perhaps aims at reducing the cost of goods consumed by the current ‘haves’. The needs of the have not-s may be very different. They may not be provided now as they are costly. So some thing not made available today, because it is costly, is not going to find a low cost automated solution.
Developed nations are the orderly side of the world. The other side of the world is getting out of control. May be we should start talking of ‘inclusive automation’. We should not look at basic minimum wages being given for free. That is not sustainable. Prices will rise when people have an income without having to do work. The starting point can be helping the poor to acquire skills for their survival. How do we automate this skill
imparting process?
Thanks for the, yet again, very thoughtful comment. And thanks for sticking in there when you were having problems with the commenting function here, Sowmyan.
You are right. There is a kind of “code within the code” that drives most automation (and the machine learning that is increasingly driving it). That code is maximizing profits. I don’t think there is anything wrong with profit, mind you. It’s more a question of what we do with those profits. If we simply extract them out of the organization, then we are essentially extracting them from all the stakeholders who were critical to getting the organization where it is now — and essential for getting it where it needs to go next.
That’s tied in very much with your idea of “inclusive automation,” I think. That term connotes stakeholders to me. I’m curious what kind of skills you see most needing to be imparted.