The question of how artificial intelligence (AI) will impact human well-being is something of a Rorschach test. It depends on whom you ask. For some, the advent of “smart” everything portends a future of rising living standards and unimagined opportunities for human flourishing. Others help us see the dark side. Without going “full Terminator,” some of the best informed people in the AI field see an economic transformation that threatens social cohesion as human labor — mental and physical — becomes slow, antiquated and artisanal and is ultimately consolidated and replaced by the unlimited energy, dynamism, and rocketing productivity of an algorithm-driven world.
Two recent books, Work in the Age of Robots (Encounter Books, 2018) by the Manhattan Institute’s Mark Mills, and AI Superpowers: China, Silicon Valley and the New World Order (Houghton Mifflin Harcourt, 2018) by Kai-Fu Lee, former head of Google China, embody the polarities of this debate. For Mills, the supposed threat of AI to human work is same macro-economic mistake we’ve been re-living since the Kennedy Administration: we see automation destroying jobs but can’t yet envision the economic demand and jobs automation will create. For Lee, this chapter of innovation is truly different. AI is like the introduction of electricity, a “breakthrough technology that once harnessed can be applied to revolutionizing dozens of different industries” so fast and so disruptively that it will result in massive economic dislocations, and overwhelm the ability of individuals and societies to adapt. Mills is sanguine; Lee sanguinary.
Lee’s argument is rooted in his career as an architect and builder of the modern Chinese economy, his stewardship of Google’s failed entrance into the Chinese market, and — surprisingly — his battle with lymphoma. He describes the Chinese economy as being built on what he euphemistically calls “imitation” — known as “theft” in the West — of technology chiefly from North American innovators. This technology has been assimilated and is morphing under the pressure of a gladiator entrepreneurial culture of rapid-cycle competition and development. China has taken what it needs from the West and is now leaping ahead by applying that technology to products like WeChat, a “Swiss army knife” app that combines social networking with everything from grocery shopping to booking doctor appointments and airplane tickets. Backed by oceans of data (Lee calls China the Saudi Arabia of consumer information) Lee predicts Chinese firms will garner more than half the projected $15 trillion increase in global GDP that AI will generate.
In the world Lee envisions, the money gleaned from AI innovations leads not to rising living standards but to economic and employment dystopia. Vast wealth accumulates to a small number of early AI investors and developers while the rest of humanity scrapes together a bare subsistence performing tasks requiring too much manual dexterity for robots. In Lee’s view, AI is a general purpose technology, like steam engines, that reshapes all production. Unlike the Industrial Revolution, where the introduction of industrial-scale mills provided employment to millions while wiping out the jobs of a comparative handful of small artisan weavers, AI will do the opposite. Rather than breaking work down into assembly-line processes that require more workers, AI eliminates jobs by concentrating work in the hands of a few highly skilled technology workers managing algorithm-driven systems. Eliminate hundreds of thousands of skilled tax preparers and replace them with the effectively unlimited capacity of web-based Turbo Tax. In other words, AI takes aim at the cognitive workers who do lots of repetitive “assembly-line” information tasks of the types that provide a lot of middle-skill, well-paid work. The robots are coming for the factory workers, and for the information economy professionals.
To Mills, the Lee scenario is amped up Luddism. Every human being, he says, is allotted about one million hours in a normal life-span. The story of human advancement has been one of figuring out how to substitute machine for human labor thereby reducing the number of hours devoted to backbreaking drudgery. AI is merely the next step in that journey of reducing labor inputs and increasing productivity, wealth, and well-being. In this view, economic growth is centered in manufacturing but surrounded, as it has been since the dawn of the modern era, by a much larger service economy. Demand for new and existing products is nowhere near its limit. In the developed world, there are 700 cars for every 1,000 people; in the developing world there are several hundred people for every car. If AI boosts productivity and reduces the price of finished products, demand will rise and with it the total employment required to supply materials, fabricate products, and distribute, sell, and service the gadgets.
And here’s where the Lee and Mills perspectives partially converge. Whatever else AI is, it cannot replace the human beings at the center of economic life. Lee learned this during his bout with cancer. Machines diagnosed his illness and he benefited from the best drugs and medical protocols. But it was his family, whom he says he neglected during his career, who really got him through treatment. They paid exquisitely focused attention to his needs — down to observing his reactions to the food they prepared and adjusting the spices to reduce nausea — something a machine could never do. Lee’s conclusion is that we can employ those who might have worked in factories and offices in jobs that provide education and care. State subsidies, paid for by taxes on the tidal wave of AI generated wealth, would raise compensation for those jobs to a respectable level. Similarly, Mills notes that it is absurd to assume that machines, however sophisticated, could ever replace people. An internal Google study listed eight hiring qualifications required by managers. At the top were communication, cooperation, and critical thinking — distinctively human characteristics. At the bottom were degrees in hard sciences.
The human element both Lee and Mills emphasize is the bedrock on which the economic future is built, regardless of which perspective ultimately turns out to be more accurate. The sci-fi nightmare of a general artificial intelligence capable of replicating or imitating human behavior is a mirage. Algorithms operate in narrow, well-defined tasks. The human mind is like an infinite number of highly complex and perfectly synchronized and integrated algorithms that are built, first and foremost, to respond to other human minds. This inherent human sociability, which Adam Smith identified as the source and purpose of economic life, is what will shape our future not the silica pathways of our magnificent computing machines.
Brent Orrell is a resident fellow at the American Enterprise Institute. He previously worked for the legislative and executive branches of the U.S. government.