The problem of the machine simply won’t go away.
Even the Bank of America Merrill Lynch (pdf), while identifying the potential benefits (to consumers, based on lower prices, and some businesses, at least the first adopters) of the “creative disruption” occasioned by current forms of technological innovation, spends a good bit of time examining the possibility of rising inequality.
One of the great concerns of innovation is the potential disruptive effect upon the labor market. “Technological unemployment” is a long-held fear that is more relevant for certain individuals than whole economies – at least for now. The greater challenge is how creative disruption can give rise to winner-take-all and monopolistic outcomes. These can actually create incentives for entrenched incumbents to spend more effectively defending their monopoly rents than to innovate further: consider Microsoft’s defense of its Windows operation system near-monopoly for a time. Similarly, the first to market may benefit from sizable first-mover advantages that create strong network effects for the first, rather than the best, technology. In addition, digital innovations create much larger reach for any given entrepreneur, as near-zero marginal cost allows firms to scale up easily. All of these trends tend to concentrate market power and wealth, and thus can exacerbate trends toward greater inequality.
In addition, skill-biased technological change rewards the highly educated and highly skilled over others. More recently, innovative uses of data collection, processing and automation have reached well beyond the factory floor: bank tellers, x-ray technicians, paralegals, secretaries, and many other service positions that once were middle-skill and middle income have been disappearing to the relentless rise of innovation. It may be only a matter of time before jobs we now consider higher skill and higher wage are similarly replaced. As just one example, sophisticated automated systems for wealth management are already under development. Like so many digital services, these have low marginal costs and scale easily, resulting in much lower costs to produce and thus prices for consumers – but also fewer opportunities for employees.
The limiting case here would be general purpose robots that are effective substitutes for human labor but at a fraction of the cost. In that case, widespread unemployment could be an outcome – it depends on whether there develops a large enough sector in the economy where humans have a comparative advantage. This could be the arts and entertainment, or personal care services, or areas that involve deeper analytical thinking that is not amenable to existing forms of AI. The transitions from agriculture to manufacturing, and then manufacturing to services, were feared by some to result in mass unemployment. What happened instead is that some old jobs gradually disappeared as technological progress supplanted them, while new – often unanticipated – jobs arose in their place. This was not always ideal for individual workers, who may have found it very difficult or near impossible to make the kind of transitions needed to gain new work, but overall neither of these transitions caused a massive rise in unemployment. The same may well be true for the next transition.
It may—but I’m not holding my breath.
It is, of course, the case that workers’ wages depend on the number of workers looking for jobs and the rate of growth of employment opportunities, which in turn depend on both the degree of labor substitution in employers’ adoption of the new technologies and the overall rate of economic growth.
Slower expected growth in the years ahead, accompanied by corporations’ decisions to automate many production tasks (of both goods and services), represents a menacing prospect for the majority of workers. They will be adversely affected both by real technological unemployment and by the threat of technological unemployment.
One possibility is to worry about and search for measures to raise the rate of economic growth, so that displaced workers have a higher likelihood of finding jobs in new, growing sectors of the economy. Fast growth is unlikely but that’s what mainstream economists are focusing on today. The other possibility is to question the nature of the new technologies that are being adopted—to challenge not technology per se but, as I have argued before, capitalist technology.
The fact is, in an economy characterized by obscene levels of inequality in the distribution of income and wealth, the adoption of new technologies is guided by those inequalities—and is likely to make them even worse.
And that’s exactly what worries Stephen Hawking.
*But, as always, there’s Matthew Yglesias who attempts to argue that the problem is not automation, but the slowdown in the pace of productivity growth.