If you’re like me, and in search of a much-needed reprieve from the professed doom and gloom of a hyper-autonomous and apparently jobless future, then I have just the book for you. Our Robots, Ourselves: Robotics and the Myths of Autonomy, out last fall, takes a hard look at the realities of the coming wave of automating technologies, such as advanced robotics and artificial intelligence, and the role humans will play in an increasingly digital future.
The book’s author, David Mindell, is the Frances and David Dibner Professor of the History of Engineering and Manufacturing and a Professor of Aeronautics and Astronautics at MIT. In the book, he applies decades of experience working at the forefront of advanced automation as a guide for how use-cases of these technologies are likely unfold as they make inroads into a broad set of industries.
Mindell throws a healthy dose of cold water on wild predictions about a jobless world overrun by robots. His argument hangs on two major points. First, human and robot interaction has been, and always will be, vital to automated systems. Much like the computer revolution, the future of intelligent automation will deeply involve the interaction of man and machine—from concept and design, all the way down to users. To Mindell, many commentators and newcomers to the discussion miss this entirely—giving rise to what I view are hyperbolic fears of a humanless workforce.
Secondly, if you want to know where the future of automation is headed, look no further than the application of these technologies in extreme environments today. At the technological frontier, the use of robotics and artificial intelligence in areas like deep sea, aviation, war zones, and space, hold clues for how these technologies will eventually be absorbed into areas like health care, finance, and education. In each of these extreme environments, humans are intimately involved with machines—again, from beginning to middle to end.
For Mindell, the key issues do not concern manned versus unmanned, nor human-controlled versus autonomous—these miss the fundamental principle that the human footprint is inextricably linked with modern automation. Instead, more appropriate questions include: Where are the people? Which people are they? What are they doing? When are they doing it?
Mindell points to three mythologies of robotics and automation that have perpetuated these misunderstandings:
· The myth of linear progress – “the idea that technology evolves from direct human involvement to remote presence and then to fully autonomous robots.”
· The myth of replacement – “the idea that the machines take over human jobs, one for one... rarely does automation simply ‘mechanize’ a human task; rather, it tends to make the task more complex, often increasing the workload (or shifting it around).”
· The myth of full autonomy – “the utopian idea that robots, today or in the future, can operate entirely on their own… automation changes the type of human involvement required and transforms but does not eliminate it.”
A well-known application of autonomous technology that Mindell says violates all three of these myths is none other than the Google driverless car. In a series of carefully crafted public pronouncements, shifting forecasts, and strategic resets, Google has decided to remove the interaction between man and machine altogether. For Mindell, this serves as both a manifestation of how difficult it is to design a human-centric, semi-autonomous machine, and a strategic blunder on behalf of the Google team.
Like other semi-autonomous systems operating in extreme environments explored throughout the book, the driverless car will be faced with system failures, variability of skills among users, problems of attention management, the degradation of manual skills, and rising automation bias as people come to rely on automated systems. Mindell suggests this is likely why Google has shifted from a model where human drivers can be in control, to one that has no driving wheel or console.
In short, Mindell believes that by avoiding one set of difficult but probably solvable challenges (human-machine symbiosis), Google is creating a series of potentially insurmountable ones (by failing to understand the three myths of automation). I lack the expertise to assess the validity of such statements, and am not oblivious to the fact that they are highly controversial. But, I do see the limitations of viewing these advances as purely abstracted technical problems to be handled with brute force engineering, rather than within a broader human or societal context. Furthermore, I would say that someone with David’s track record suggests his words should be considered seriously.
Overall, I greatly appreciate having read this book. As a non-technologist, it gave me a clearer picture of the technological possibilities and challenges associated with the future of automation, and some big ideas about where things are going. Perhaps I also liked it because it reconfirmed some of my priors about the inherent limits of automation.
To be clear, I am a technology optimist. I believe that we are on the cusp of a wave of advances in digital technologies that will fuel economic growth and improve our quality of life. However, I also know that nothing about our past or what we know about our present suggests the dystopian future that some have envisioned. Humans and machines will always co-exist, information will be one of many inputs to production (and consumption), and a sizable gap between technological possibilities and technological absorption is certain.
Regardless of your views on these latter points, I recommend this book to thinkers who are looking for a reasoned assessment from a seasoned technologist about where automation has been and where it’s going next. For those of you looking for a quicker summary of content, see this great podcast where Mindell appears on EconTalk with host Russ Roberts of the Hoover Institution. Enjoy.