It seems there is almost a daily story on the effect robots and automation will have on the current labor force. Autonomous, self-driving cars and trucks, robot mops and automated pizza delivery vans are at the horizon’s edge of a future economy that promises to redefine the interplay between humans and machines in the production of work.

Estimates indicate 47 percent of current employment in the United States has the potential to be automated in the next 10 to 20 years based on current technology trends. However, potential risk is not the same thing as inevitable replacement, and research shows that while some jobs will likely be fully automated, most will be redefined as automated systems and robots are introduced into the economy.

In the past, automation in the workplace was limited to manually routine jobs such as in agricultural and manufacturing, and cognitive routine jobs such as office tasks of processing forms and bookkeeping. Now however, machine and computer engineering has begun to allow automation of jobs that are both physically and cognitively non-routine. This is why 47 percent of employment has the current potential to be automated.

But what are these jobs and what are the other 53 percent of employment?  How easily can they be automated? These figures are from the 2013 Oxford report “The Future of Employment” by researchers Carl Frey and Michael Osborne. They identify intrinsic human traits such as social intelligence, creativity, perception and manipulation as bottlenecks of current development for engineering machines to perform workplace tasks. Jobs that require a high degree of one or more of these identified human traits are less susceptible, or at risk, of becoming automated because developing the technological capability for machines to replicate these traits in place of a human is not yet possible. Jobs that use these traits to a high degree and are less susceptible to automating include such jobs as surgeons, therapists or public relations professionals.

Inversely, jobs such as dishwashers, telemarketers and retail salespeople, do not require workers to have high proficiency in any of the bottleneck traits and are more susceptible to automation. Applying this methodology to 702 occupations in the United States, the research team developed a probability that each of these occupations were susceptible to automation.

Fig. 1 is a replication of the chart Frey and Osborne developed to show the results using Idaho data rather than national data.

The proportion of Idaho’s employment at high risk to automation is similar to the nation at 47 percent. In broad occupational category terms, (Fig. 2), Idaho employment in occupations requiring advanced technical skills and interactions with other people as part of their work are at the lowest risk.

Among these include the computer, engineer and science group or STEM occupations.  Fig. 3 shows that 80 percent of STEM employment is low risk and not likely to be automated in the foreseeable future. Employment with a larger share of jobs with high susceptibility to automation include construction, sales, service and supply chain occupations. Production, office and administrative support, and farming, fishing and forestry not only share that high-risk status, but also provide the historical context that what we are experiencing is not entirely new.

Automation of the workforce we experience today is a new phase in a very old process. In the last century, machines have replaced human hands performing routine tasks in the fields, in factories and in offices – in each case to improve productivity, increase quality and consistency while reducing error. Automation of agricultural occupations transformed the industry accounting for a drop from 41 percent of the workforce in the United States in 1900 to just 2 percent by 2000 (Autor, 2015). “Whether the technology is tractors, assembly lines or spreadsheets, the first-order goal is to substitute mechanical power for human musculature, machine-consistency for human handiwork and digital calculation for slow and error-prone ‘wetware’ (i.e., brain power).”

If history is a good example then the implication is that human activity will be complemented, rather than replaced outright. Accountants and auditors from a labor market perspective are good occupations as they are in strong demand, plentiful and pay good wages. However, in Frey and Osborne’s methodology there is a 94 percent probability of those jobs being automated. Does this mean in  20 years artificially intelligent machines take care of all our accounting and auditing needs? It is unlikely that human accountants will be go the way of the buggy-whip maker. Rather, a considerable portion of their tasks will be automated, their efforts will be complemented and the occupation will likely be redefined rather than eliminated (Chui, Manyika, & Miremadi, 2015). Accountants and auditors are not unique in this situation as up to 45 percent of work activities performed by individuals can be automated across the spectrum of occupations, even for those at low risk of automation such as surgeons, attorneys and CEOs.

Predicting the long-term future in any context is generally a fool’s errand, and in no way should information presented in this article be taken as a prediction. However, the implications of the trends in machine robotics and machine intelligence on the labor market is that machines performing in the place of people will become much more ubiquitous and likely expand our capability and economy as a whole rather than a portend of doom.

Craig.Shaul@labor.idaho.gov
regional economist supervisor
Idaho Department of Labor
(208) 332-3570 ext. 3201