At the end of the 18th century, Samuel Slater brought manufacturing technologies from Great Britain and built the first American cotton mill in Beverley, Massachusetts, signaling the start of the Industrial Revolution in the United States. Over the next one hundred years, the Industrial Revolution fundamentally and permanently changed the American social and economic landscape. Now, just over two hundred years after Samuel Slater first arrived in Massachusetts, the United States and the world are at the precipice of another technological revolution: advanced and wide-ranging machine-intelligent automation. This machine intelligence revolution will give rise to new social challenges in preparing for and adapting to an automated future.
Automation in some form has been around since the industrial revolution, but recent advances in machine intelligence coupled with falling prices of machine intelligent technology have accelerated the prevalence of automation in a growing number of industries. Boston Consulting Group estimates that global spending on robots will jump from $15 billion in 2010 to $67 billion by 2025. In addition to the steep upward trend in investment into machine-intelligent technology, the pace of technological innovation itself is astounding. In 2004, economists Frank Levy and Richard Murnane wrote a book entitled ‘The New Division of Labor,’ where they argued that there are limits to the capabilities of machine intelligence, writing that “executing a left turn against oncoming traffic involves so many factors that it is hard to imagine discovering the set of rules that can replicate a driver’s behavior.” However, by 2012, Google’s self-driving cars had driven over 300,000 miles.
Machine intelligence is not only slated to change the way that we travel; it will more importantly change how we work, where we work, and what we work on. Previously, demand for intelligent robots has been highest in high-wage manufacturing sectors, such as the automotive industry, but demand has begun to diversify to take on tasks outside of routine manufacturing. Such tasks now include functions that used to be seen as uniquely human skills, such as induction, pattern matching, natural language processing, image recognition and categorization. In 2015, Microsoft, working with the China University of Science and Technology, announced that their artificial intelligence software achieved IQ scores at the college post-graduate level.
As the abilities of machine intelligent technologies increase, the more they have the potential to do jobs otherwise reserved for humans. Pew Research Center writes that “Certain highly-skilled workers will succeed wildly in this new environment—but far more may be displaced into lower paying service industry jobs at best, or permanent unemployment at worst.” In order to quantify the effect of advances in machine intelligence on the United States workforce, two Oxford researchers, Carl Frey and Michael Osborne, composed a model in which they separated jobs into three categories: those with high risk of being automated within the next two decades, those with a medium risk of being automated, and those with a low risk of being automated. According to their model, 47% of total U.S employment falls into the high-risk category.
A common response to the notion of future job replacement by means of automation is that while jobs of today may be replaced, the technological increase will cause economic growth and create jobs in other sectors. However, machine intelligent technologies are unique; as former Treasury Secretary Lawrence Summers explains that “today, there are more sectors losing jobs than creating jobs. And the general-purpose aspect of software technology means that even the industries and jobs that it creates are not forever.” As a result, we can expect see a mass of intermittent employment, as many working people will be forced to switch jobs often.
Additionally, this wave of automation will cause rampant growth in income inequality, as those who own the technology will be set to make millions, while others will struggle with unemployment or underemployment. Machine intelligence has already begun to result in this labor market polarization, where employment is split between high income cognitive jobs and low income jobs, while routine middle income jobs are being hollowed out.
This dramatic reshaping of the United States workforce will also require a reshaping of social policy in the United States. Darrell West, founding director of the Center for Technology Innovation at the Brookings Institute, writes “if we end up in a situation with many people unemployed or underemployed for significant periods of time, we need a way to provide health care, disability, and pension benefits outside of employment.” West suggests that one way to do this is through lowering the number of hours that people have to work in order to qualify for employment benefits. While this idea would help the growing number of underemployed workers in an increasingly automated economy, it would be unattractive to employers by making them pay for more benefits, while not providing enough support to the unemployed.
Structurally, the simplest social policy would be a guarantee of a Universal Basic Income, or UBI. The idea of UBI gained modest notoriety in the United States when it was endorsed by conservative economist Milton Friedman, who argued that it was a way to fix the inefficiencies of the welfare bureaucracy. The goal of UBI is to provide people with a small, unconditional monthly income that covers necessities. Last year Sweden voted on, but ultimately rejected, a referendum that would supply adult citizens monthly with the equivalent of $2,555, with an additional $642 per child.
There are generally two main arguments against programs involving a UBI: it is expensive and it decreases incentives for work. As for the cost of UBI, depending on the amount of subsequent cuts to Medicare, Medicaid, and Social Security, UBI has the potential to actually save money over the current system. In terms of the incentive structure of UBI, Charles Kenny at The Atlantic writes that when the U.S experimented with a “negative income tax” between 1968 and 1980, these cash payments “were associated with reduced child malnutrition, improved school attendance, and growth in household assets. The cash transfers also had significant effects on children’s test scores.” Kenny added that there was only a “small decline in household working hours among beneficiaries, [and] this occurred primarily among second-and third-earners in a family rather than the primary (usually male) worker.” Not only would a Universal Basic Income policy help keep people who are underemployed or between jobs afloat, it would also help with educational opportunities for children to compete in the new job world.
The development of highly equipped machine intelligence has the potential to make immensely positive impacts on people’s lives, including advances in medicine and the spread of life-saving technologies to developing countries. Technologies may also emerge that will benefit society in ways we cannot currently imagine. Regardless of how positive the effects of widespread machine intelligence may be, the effects of the incoming massive shift toward machine intelligence are likely to shake the job market down to its core, necessitating the need for a fundamental social safety net for those affected in the form of a Universal Basic Income.