A representative from a major architectural software firm gave an interactive presentation here at Carnegie Mellon recently, then opened the floor up to questions from the student audience. One student, recognizing the way in which new software greatly increases the productivity of a single architect, brought up the elephant in the room: doesn't this software contribute to even greater unemployment by enabling fewer architects to do more of the work? The speaker acknowledged this risk, and brought up a statistic I had not heard: following the 2008 real estate crash, thirty percent of the architects in New York City have not gone back to work. This number is frightening, and echoes a widespread trend that is identified in Erik Brynjolfsson and Andrew McAfee's excellent double-header, Race Against the Machine and The Second Machine Age. The onward march of technology innovation is constantly shifting the boundary between work that machines can affordably take on and work that still remains in the human province. This boundary moves faster every year, and only moves in one direction: to the advantage of automation and a shift of income for laborers to the owners of capital. What's more, every recession and every crash provides more fuel for the labor-to-capital shift, compelling labor layoffs that are followed right quick by compensatory machine innovation to keep the company afloat with fewer employees. And when that crash ends, no company ever trades its newfound productivity to scale back automation and hire more humans.
As Brynjolfsson and McAfee put it, all routine physical jobs and all routine mental jobs are at risk. And while earlier works, including Race Against the Machine, argued that human ingenuity and innovation will surely compensate by inventing new job categories, even their second book hints that the rate of automation's advance is easily outstripping society's ability to innovate new job categories. Ironically, what few routine jobs are left often become nightmares along the path to human automation. Simon Head, in Mindless, tells the story of how humans at Amazon's warehouses are tracked and trained to move their bodies just so, shaving seconds off of product retrieval times by evaluating exactly how legs, torsos and arms move in harmony. We aren't simply replacing humans with robots; in some cases we are now training humans to act like robots.
In every recent interview I am asked the same question: what should we do about a trend that edges us toward chronic underemployment every day? There are the Age of Plenty adherents, believing that the future "zero marginal cost society" will give all of us, even the unemployed amongst us, all the nearly free goods we want (so long as we listen to advertisements bundled with the product). Imagine a life of pareto-luxury with no need to actually get out of bed and work. Then there is Piketty's proposal: a global wealth tax to support a new social contract. Finally there is the popular notion on the blogosphere of joint capital ownership: automation technologies should be owned fractionally by all the workers whose jobs are displaced by that new capital. My concern with all of these approaches is that they fail to address the underlying issue that relates directly to our sense of meaning and purpose: the loss of labor itself. The labor share of GDP is falling off a cliff, and the corporate profit share of GDP continues its ascent. We live in a time when labor power is cavitating and capital power is on the rise. This is very bad news: capital ownership is the most unequal of all forms of productivity, making the inequality in labor income pale in comparison.
So, how do we give labor a fighting chance? Take a look at societies that explicitly value service and labor. Watch retail interactions in Japanese stores, and take a trip to a gas station in South Africa. These are places where far more people are employed than absolutely necessary, all in the name of high-touch service. Our U.S. automated supermarket checkout lines, driverless taxi projects and zero-worker construction site projects fly directly in the face of privileging human service. This has to change, and that change is not impossible to imagine. Consider the driverless car trope, and tack instead toward the autopilot concept espoused by Tesla Motors. Stop removing the driver; instead create intelligent automotive features that make the human driver safer. Is a driverless New York cab really cheaper than one with a cabbie? The existing maths that have been reported are dead wrong, as I have written, and that does not even begin to take the externalities into account. A global wealth tax is unlikely today, and it is becoming clear that innovation will not save us this time, which brings me to the title of this piece.
I would throw another idea into the cauldron, based literally on computing the external cost of automation in terms of labor income offset. Let's compute the labor income cost of automation in products and services, then add that duty to the market cost of the product, just like tax. If we begin to even out the cost differential between automation and high-touch service, then we will empower businesses to seriously consider the human-automation trade in both directions, for the first time. This will never succeed unless we make the decision economically symmetric, and that is a price we can choose to pay if we wish to preserve a middle class option for our grandchildren.
Technological progress has led to incalculable increases in quality of life, in health and nutrition. But the "progress" enjoyed by the many does not extend into our presently laid future trajectory. Increasing productivity today benefits the margin of society, and waiting to see if the dynamics of employment self-correct is no good. We have a duty to the middle class, and it is time to push our politicians into actively engaging on an issue that ought to be front and center. The automation fringe is not going to threaten politicians' jobs until the very end, so we must engage with fallible, human legislators and stop waiting for superior robo-senators to replace them.