What If Robots Take Your Job: Mass Automation and Unemployment
By Lang MingPublished April 30, 2018"Okay Google, what is a big policy issue in America?"
Google Now, Amazon Echo, iOS Siri—artificial intelligence, or AI, can be seen almost everywhere in an average American's life. While machines used to be more efficient than humans only in certain tasks, more recent evidence suggests that if the technology is mature, computers can virtually do anything humans are capable of and become an ideal source of labor to replace humans due to its high efficiency and low cost. Researchers found that in 2013, 47% of workers in America had jobs at high risk of potential automation. Some may claim that automation has already been happening for decades and we never really were in trouble because of it, but the current technological advancements happening are nothing like what we have seen. The U.S. government currently has little regulations on automation, which should be concerning to everyone, including engineers and those who work with technology for a living.
One of the two key distinctions between the current and previous automation is the variety of occupations impacted. Prior to 2005, physical labor has been the primary target of automation; people who worked around construction sites and assembly lines were replaced by machines. In contrast, the workforce that predominantly relies on mental labor was not negatively affected; it actually grew as a result of automation, since the overall productivity of labor increased as production became more efficient. The automation we face currently, however, replaces mental labor as well. For instance, driving is considered to be mental labor since it requires the driver to make judgements and react based on road conditions. Although no one in the 90s could have imagined that automation would replace drivers, as self-driving car technology matures in the next several years, the time left before human drivers lose their jobs may be minimal. When machines substituted workers who provided physical labor, we moved to mental labor instead; with mental labor now replaced, what will we have left?
Another distinction between the traditional and current automation is the amount of time required for the automation to happen. In the past, technology could only replace labor at a slow rate because computer just started popularizing. It took companies years or even decades to implement automated systems. Since the process happened gradually, workers had enough time to find new jobs and mitigate unemployment. However, the situation is no longer so optimistic nowadays. Technology grows exponentially, meaning the speed its growth rate increases every year. Thirty years ago, Internet just started rising around the globe; today, we have AIs like AlphaGo beating the best chess and Go players in the world. Given how technology in the U.S. is already highly integrated into both industrial and domestic activities, once the technology for certain careers matures, mass automation can become a nightmare. By the end of 2016, 40% of the American employment only happens in four criteria: office and administrative support, sales, food preparation and serving, and transportation and material moving. However, the technology to replace every one of them already exists. Smart office technology can automate workspace; self-serving grocery stores can automate sales related jobs; AI chef has already entered the kitchen; self-driving cars will soon replace human drivers. Mass automation is coming, and it can happen overnight. Without proper regulations, we are not prepared to enter the whole new world of technology.
One potential solution to this situation is to tax companies based on the amount of automation they use and invest such revenue into improving our welfare system. Since mass automation does boost the productivity of our economy, an effective distribution of resources will allow people to keep living an adequate life even when mass unemployment hits. This is especially powerful in addressing the arguments from supporters of automation without regulations, most of which claim that government should not intervene the process of automation because instead of unemployment, automation creates jobs. According to them, two types of jobs are created: some already exist, because increased productivity leads to increased demand, the jobs related to the same product also demand more labor; some are new, such as the management of machines. However, the first part of the argument falls flat given the extreme automation that could happen based on the status quo. Demand cannot increase infinitely; when it reaches the point of diminishing marginal return and yet almost all the jobs are automated, a spike in unemployment is inevitable. The second part of the argument seems to be true, but it fails to address the classicism it perpetuates: the jobs survive or created by automation require high skill levels or high level of education, which comes from certain resources that those of lower socioeconomic statuses cannot access. As a result, the increased overall capital brought by increased productivity ends up accumulating in the pockets of the rich, which worsens the gap between the wealthy and the poor. By taxing automation and utilizing the money to improve social welfare programs, a more equitable redistribution of the increased capital allows everyone to enjoy the benefit of technological advancements. Government becomes the new "employer" of its citizens to combat unemployment brought by mass automation, while everyone is still able to enjoy an increased quality of life.
There is nothing wrong with loving technology and supporting its advancement. Automation is not undesirable in of itself; it is the lack of regulations and planning from the government that induces its troublesome side. With proper policies, technology will soon be truly life changing—and in a good way.
"Okay Google, when will America regulate automation?"