Last week, New York Times journalist Eduardo Porter posted an outstanding article on future chronic underemployment stemming from our advancing robot/AI future: Tech Leaps, Job Losses and Rising Inequality. Coincidentally, I just visited beautiful Cornell University last week to give a public Bovay lecture on exactly this topic, and so recent research gives me confidence to strongly endorse many of Porter’s theses. He points out that structural changes threaten prior automation-displacement dynamics, in that knowledge and skills opportunities are no longer increasing anywhere near apace with the erasure of formerly human labor categories. A particularly compelling witness to the new trends come to us by way of Nobel prizewinner Robert Solow, who once argued that growth effectively offsets shrinking labor categories due to innovation and productivity increases. Now, Dr. Solow says, is no longer quite to similar to then:
He cites “everyday reasons,” including the erosion of the minimum wage, the decimation of trade unions and anti-labor legislation.
But technology clearly plays a role. “We will know better in 10 or 15 years,” Professor Solow said. “But if I had to interpret the data now, I would guess that as the economy becomes more capital intensive, capital’s share of income will rise.”
As Piketty eloquently points out, capital inequality far surpasses labor income inequality (which is already terribly high in the U.S.!), and the extrapolation that I proposed at Cornell is as follows. Everything we know about our robot future suggests that robots are on the verge of becoming very efficient mining tools, extracting value from labor and refining the value into capital income. The result will be an accelerating shift in the balance of wealth from labor income to capital income- an inexorable march toward inequality, underemployment and ever-decreasing labor growth. These are very, very early days yet.
The Economist just published a special report, Rise of the Robots, that attempts to outline major trends in robotics research and the ramifications on the society of the future. In Immigrants from the Future, Oliver Morton correctly argues that robot designers and marketers do not yet understand the emotional consequences of robots in our midst– I will second that thesis quickly. In one of the richer sections, Good and Ready, Morton reports on the entrepreneurial activity in robotics, and points out just how Google and other information-age companies have a major role in the future robots that will pervade society: robots need information about all of us, but also about all things prosaic, like how to operate various human-centered designs such as locks and doorknobs. To the rescue shall come the Cloud, with services for robot recognition, interaction and operation. Robo-Google, as I called it in Robot Futures, is now part of the common parlance of robotics’ future.
There is one particularly egregious mistake that I feel compelled to point out in this broad report, and this relates to the military robotics section, Up in the air. Morton argues that robotic automation is essential to more effective war-fighting machines in some regimes, and he makes a certain confident claim:
..and for some purposes, such as defending ships against missile attacks, autonomous systems are both necessary (because of their speed of response) and legally and morally unproblematic, since they operate in areas where no civilians, and possibly no enemy compatants either, will be affected
This is a remarkable comment. Is Morton really claiming that autonomous, lethal, defensive systems don’t kill people? And that civilians aren’t endangered in such theatres of war? He may be too young to recall the USS Vincennes’ downing of an Iranian airliner, and he may not appreciate that the Aegis Combat Systems was a clear predecessor to the future’s even more automated defense systems.
Lethal robot systems will be just that- lethal. To argue otherwise, and suggest that they won’t hurt people who are innocent, is to ignore centuries of wartime statistics.
I have a proposed weekend assignment: get your self a copy of Kurt Vonnegut’s Player Piano (1952) and Simon Head’s Mindless (2014) and read them Saturday and Sunday. I just finished both, and they are outstanding bedfellows. As you read Vonnegut, you will find yourself constantly amazed at what he imagined more than sixty years ago, in forecasting a society where automation has quite literally broken the backs of the human race, removing much of craft and authentic work from the equation. Vonnegut describes a world in which the winners are an engineering-management upper class who oversee the Machines that do everything. The economic gap has widened to the point of infinity, with beautifully crafted justifications by the upper class of how the Machines have improved the lot of the massive, underemployed class– never mind they only have a few dollars to spend by choice, as what they truly need is supplied by social services in just the right amounts at the right times, automatically. Machines decide what to build, how much to build, who to hire and who to promote. They are the cognitive orthotic that dumbs down society but keeps it running efficiently and incredibly unequally, and of course the protagonist attempts to revolt, lamely, against this age of rational efficiency. The book brims with indices and quotients- with the idea that numerical estimates run everything, from what color book-jackets should be to just what authors are published (only if their readability quotient is well above 27). Remember my Robot Futures story about the color blue at Google? Vonnegut predicted this more than half a century before it happened.
Then take on Simon Head’s Mindless, Why Smarter Machines are Making Dumber Humans. I readily admit that Head is not as gifted at prose as Vonnegut, but that’s alright. Head dives deep into just how computer business systems have changed industry- replacing autonomy and master craftsman empowerment at the level of the individual worker with machine-optimized peak efficiency through micro-managing every detailed move a worker makes. His description of both Walmart and Amazon in this respect is horrifying, demonstrating how literally the hand motions of each restocker and shipper are measured, optimized and adjusted, with quantitative performance indicators that ratchet up inexorably until the older employees have special “efficiency review” failures and inevitably lose their jobs. His book sharply describes a here-and-now, in the U.S. and in China both, where the worker class are almost literally automatons– robots– and where middle management cavitates, to be replaced by machine optimization and A.I. that serves top management and top executives directly.
Both of the stories are fundamentally about the dehumanization that machine efficiency can bring, and the carefree feelings we can nurture while reading a work of fiction by Vonnegut are dashed when Head brings forward specific cases that are very much non-fiction. Come to think of it, you may want to save this double-header for a rainy, foggy, cheerless couple of days.
Thanks to Randy Sargent for sending on a link to Annie Lowrey’s article in the Times Magazine, Hey Robot, Which Cat is Cuter? This article mentions Jaron Lanier’s book as well as Race Against the Machine, both books I have references frequently in this blog, to tie together arguments regarding underemployment. The crux of the debate rests on interpreting just how job categories and needs change as automation and AI proceed on a one-way march: if new innovation forges new industries rapidly, then we humans blossom into new styles of intellectual work (think App-writing?). If the new innovation matches human capabilities at an ever-accelerating pace, then we have a runaway train problem- even structural improvements that could compensate over the long-term become irrelevant as the gap widens constantly in the short term. I find particularly telling a quote Annie provides from Bill Gates:
“Software substitution, whether it’s for drivers or waiters or nurses” is coming, Bill Gates said recently at the American Enterprise Institute in Washington. “Twenty years from now, labor demand for lots of skill sets will be substantially lower. I don’t think people have that in their mental model.”
The debate goes on, but evidence continues to build. And as with all accelerating phenomena (think climate change), once we are all convinced of the causes and effects of chronic underemployment, it will likely be too late to act with patience.
Capital, Piketty’s tome, originally written in French as Le Capital au XXI siecle, is an intimidating 577 pages. Just to demonstrate the astounding level of scholarship Piketty has engaged, consider that the endnotes consume 76 pages of small font. Yet this book is a must-read. Piketty will explain how wealth is composed of income from labor and income from capital, and then he will take a turn far away from most economic analyses that I read. He believes strongly in detail, and he eschews glib rolled-up approximations like the Gini coefficient for inequality in favor of precise, temporal detail. He thrives on collecting, cleaning, presenting and interpreting data over time, from the way capital ownership has changed in the upper centile of France’s residents since 1700 on through last year. You read that right: he tracks trends and tradeoffs not over ten years or a hundred, but rather over several hundred years. And this depth of hindsight yields breathtaking results. You learn just how unbelievably unequal our society is today, and just how more divergent and unequal the United States has become over the past thirty years, even as compared to the entire, remaining first world. You learn how dangerous accumulation of capital is in the hands of the few- how it can spiral out of control, generating more unequal wealth at an ever-accelerating pace. You will also see that overall national and global growth has no chance, over the long run, of offsetting the vaporization of job categories as human labor converts to wealthy capital thanks to: drum roll…..robotics! Okay, that last part is not explicit in his analysis, but it is exactly the lens through which I encourage you to read and understand his characterization of the dynamics of wealth around the globe. Piketty has a general proposed salve for our ills, and it all depends on a global tax on all capital ownership. Of course you, Piketty and I know just how likely it is for international, broad-reaching consensus on new global taxes.Hah. But he argues convincingly that this is one of the only chances we have to save ourselves from our own economic dynamics.
The book is very dense and very long, and I have a specific recommended reading program- not something I have ever published before. Here we go:
- Introduction: read it all to learn how the data source he has helped to create, the World Top Incomes Database, is invaluable in enabling us to really track what’s happening.
- Chapter 5: The Capital Income Ratio
- Chapter 7: Inequality & Concentration
- Chapter 9: Inequality of Labor Income
- Chapter 12: Global Inequality of Wealth
Read these five chapters and you will be able to hold very strong in any dinner conversation about just where our economy has been and where it’s headed. Add a little spice by imagining gradual conversion of labor to capital via robotic automation, as I talk about in Robot Futures, and you have the beginnings of a vivid nightmare.