EDITOR’S NOTE: This is a new RealClearBooks series, RealClear Authors, in which we sit down with authors to discuss their latest work. In this first installment, Christian I. Giadolor interviews Mark Mills to discuss his new book, “Work in the Age of Robots.” Read Mills's recent RealClearBooks article adapted from the book.
RealClearBooks: Let’s start with your book title. I always thought the advent of robots would mean we wouldn’t have to work anymore.
Mark P. Mills: Well, a robot is, essentially, just a very complicated and — so far — expensive labor-saving machine. We humans have been inventing labor-saving machines like automatic washing machines and automobiles for all of modern history. And as has always been the case, machines change the nature of work, but more or less the same overall share of people end up working.
RCB: Often, people dichotomize the service economy and the knowledge economy. But you write that the knowledge-centric economy is heavily dependent on manufacturing and should remain so. Does this mean that blue-collar manufacturing jobs, especially ones that don’t require a bachelor’s degree, may still be available for my generation?
Mills: Let’s hope so since PEW research tells us that 60 percent of millennials do not have a college degree. A lot about what really constitutes a knowledge economy is relevant not just for college-educated work, but also blue-collar jobs. It’s knowledge that can be gained either on the job or in non-college programs. And, there is no artificial intelligence unless some factory somewhere manufactures the computers and produces the materials in the supply chain that make all that possible. The world manufactures a half-million tons of portable electronics each year by the way and that itself requires tens if not hundreds of millions of tons of materials to be extracted from the earth, processed and transported. That sounds a lot like a lot of blue-collar work, even when it’s enhanced by smart supply-chains and automated equipment.
RCB: Reading your work, I began contemplating how “means of management” changes might impact our federal bureaucracy, perhaps making it more effective. Do you think automation could be the answer to helping our government work better?
Mills: Hoo boy. If ever there were a sector of our economy — and government is a huge sector — that needs to become more cost-effective and less error-prone, it’s our federal bureaucracy. I am confident that artificial intelligence and social-media-type tools can improve both the effectiveness and transparency of our government. But, honestly, I think that’s going to be a heavy lift.
RCB: “Amara’s Law,” named after former Stanford Research Institute computer scientist Roy Amara holds that we tend to overestimate the effect of a technology in the short run and underestimate it in the long run. So when do you think the inflection point of artificial intelligence will arrive — if it’s not here already?
Mills: In my book, I mapped out so-called inflection points by analogizing the birth of any radically new technology as a kind of “first fission.” The first time scientists split atoms to make energy was in 1938. It took about two decades before we built the first commercial nuclear power plants. And, with nuclear energy at least, we still have yet to see an inflection point where the technology takes off to become dominant. It’s still only about 10 percent of world electricity. It’s interesting that 20 years seems to be a rather common time period from first fission to first commercial viability for radical new tech — and then another 20 years before an inflection point in which the technology starts to becomes a common, if not dominant, feature of society. That’s exactly the time periods for both cars and computers by the way: 20 plus 20 years. I think history might see first fission for AI as the 1997 chess match when IBM’s Big Blue beat Grand Master Garry Kasparov. About 20 years have passed and we’re now seeing commercially viable AI emerging. That means, if the pattern holds, the inflection when AI becomes “yuge” could still be another two decades out.
RCB: You also write about a “democratized AI”? What would this look like? And would democratizing it alleviate the kinds of fears that have been a staple of science-fiction movies from the “Terminator” series starring California’s former governor to “I, Robot”?
Mills: Who doesn’t love Terminator-type movies? As for “I, Robot,” I was an early fan of Isaac Asimov, who wrote that book in 1950. I read dozens of the nearly 500 books he wrote — as an aside, that’s pretty amazing isn’t it? Not my reading dozens, but his writing 500-plus books. I should note that Asimov’s view of robots was far from dystopian. He really saw them much as I do — or, rather, I see them as he did — as powerfully collaborative tools. What I meant by democratized AI was the inevitable evolution to AI becoming intuitive, widespread, and easy to use — much as we happily and easily use say Google maps, or Uber, both of which are essentially rudimentary AI engines. As AI gets better, and cheaper, and is accessible using natural language questions and interactions in both work or daily life, it will enhance rather than distract from many more aspects of society. It’s inevitable that one day — certainly your kids and my grandkids — won’t be able to imagine a world without AI in common use any more than we can imagine a world without electricity everywhere.
RCB: You mention that the growth of America’s regulatory state is preventing manufacturing from reaching its full potential with technology. Which regulatory practices do you think have been most stifling? And how would policies enacted by the Trump administration alleviate some of that burden?
Mills: Start with the fact that the size of the federal regulatory bureaucracy — in both manpower and spending terms — has doubled in the past decade. Regulators have a role, but we’re clearly over-regulated pretty much across the board. The industrial sector is the biggest target of all the regulation. A lot of good scholarship on this has been covered recently by the RealClearPolicy folks, with a focus on the “administrative threat,” which, by the way, is the title of another and excellent Encounter Books’ treatment of this subject by Philip Hamburger.
RCB: As a college student at a university where issues like AI and automation are on the forefront of everyone’s minds, I was surprised to read that “soft skills” such as critical thinking and cooperation remain the most desired by employers at companies like Google. What does that mean for the future of America’s workforce? Is the intense focus on STEM by some educational institutions misguided?
Mills: I do think colleges and K–12 educators are over-focused on STEM, and I say that as a physicist and former practicing engineer and scientist. It bears noting that well under 10 percent of the workforce is employed in STEM occupations even today. If we count, say, only coders, there are still more jobs for people in farming. The truth is that STEM is critical of course, but most jobs are non-STEM and the most successful people in STEM domains generally have those so-called non-STEM “soft skills” of critical thinking, creativity, and collaboration. You learn those as much if not more from non-STEM subjects like English, history, philosophy, and language. And, it’s useful to keep in mind, that one thing powerful AI will soon do is allow non-STEM-educated people to use, call it STEM-AI , to help them design things, build things, or invent things. AI will democratize much of the knowledge inherent in STEM.
RCB: Lately, I’ve been consumed by a video game called “Detroit: Become Human” that portrays a society where automated androids have effectively displaced half of America’s workforce. But the game’s portrayal of a struggling nation is what I find most striking. Even without job displacement, how do you think automation and AI will remake our society?
Mills: Game-makers and science-fiction writers always love dystopian futures. It’s a lot more fun to write about, read, and watch. If you like dystopian stories, read “Robopocalypse,” or another book called “Kill Decision.” Talk about stories with scary war-fighting robots. But in our real future — setting aside that robots, like all technologies, will indeed be used in war and will replace many kinds of work — the main impact of AI and robots will be to accelerate productivity. That has always meant creating more wealth for more people. One important and unique effect of AI in particular is that it will make more workers more knowledge-capable, or to put it another way, it will make far more people capable of performing so-called knowledge-work.
RCB: In researching “Work in the Age of Robots,” what did you learn that surprised you?
Mills: Aside from learning how astonishingly common and intransigent the belief is that robots mean the end of work, I was intrigued by how history shows my first-fission thesis to be amazingly robust — that it takes 20 years from first fission to practical products, and then another 20 years before an inflection. That was also the timeline for the internet, by the way. This flies in the face of the everything-is-happening-faster claim about technology. Yes, some things happen faster. But really foundational changes in technology and society are running at the same kind of pace they have for 150 years. That doesn’t mean, I should say, that social disruption or political impacts from the mere prospect of change doesn’t happen fast. We witness that all the time.
Mark P. Mills is a senior fellow at the Manhattan Institute and a McCormick School of Engineering Faculty Fellow at Northwestern University. His book,“Work in the Age of Robots,” was published this month by Encounter Books.
Christian I. Giadolor, a Texas native, is a RealClearPolitics summer intern. He is a student at Stanford University majoring in political science.