The Potential Rise of Robotics in High-Skill Jobs
August 25, 2017
The continuing proliferation of algorithmic decision-making, automation, and the use of artificial intelligence has once again brought to the forefront of the minds of those employed and those seeking to keep unemployment numbers down the question of the potential impact of automation and artificial intelligence on the labor market. While some are most concerned about the potential effect of job replacement and job loss thanks to automation, others are more concerned about biased decision-making coming from algorithms, a lack of quality checks for algorithms and automated systems, and the potential implications of automating jobs such as those of government employees or regulators.
The development of artificial intelligence as well as advanced robotics provides for increased efficiency with decision-making and the completion of low-skill tasks, but also poses a threat to jobs in both high-skill and low-skill areas. While automation can pose potential threats to certain jobs, increased automation leads to the creation of new, complimentary jobs, meaning that workers whose jobs may be in peril can prepare and be trained for this opportunity. A series of examples regarding the potential use of automation in jobs characterized by required high education levels, job training, and a need for proficiency in abstract thinking, have emerged in recent years, demonstrating the growing potential of automated technology.  Currently, police forces in Durham are using algorithms to decide if a suspect should be kept in custody or released on bail by classifying them as high, medium, or low risk for recidivism and not returning for their trial.  Doctors have also begun using AI to read CT scans and MRIs using deep learning, a technology demonstrated to have superhuman abilities in the areas of image recognition and accuracy.  IBM’s Watson is currently being tested to replace many jobs considered to be high-skilled such as those largely including hypothesizing, reasoning, and other skilled human tasks through cognitive automation.  In the future, it may be possible for algorithms to arbitrate decisions, decide on who should receive federal benefits, and enforce rules and regulations for fast-paced industries.
In terms of potential job replacement, predictions differ. Researchers Carl Benedikt Frey and Michael A. Osborne predict that 47% of jobs are candidates for partial or full automation. However, a competing estimate from McKinsey Global Institute states that 5% of jobs being candidates for full automation and nearly 100% of jobs being candidates for partial automation.  The Organization for Economic Cooperation and Development states that on average, across the 21 OECD member countries, 9% of jobs are able to be automated.  One of the main issues that needs to be dealt with going forward is the lack of actionable data regarding potential job replacement.
Another key issue is potential biases in decision-making and algorithms. A recent ProPublica study on algorithms that are currently being used by police to rate the probability of an individual committing a future crime has discovered widespread racial disparities in terms of risk assessments. The algorithms used wrongly labeled black defendants as future repeat offenders twice as often as white defendants and white defendants were mislabeled as low risk more often than black defendants.  While machines may be able to make more accurate predictions than humans based on the data given, unlike humans, machines are unable to self-correct or adjust for potentially biased data, leading to potential concerns when algorithms and technologies seen as unbiased are actually operating off inherently biased data. This area is one where government regulations, including mandating ways to adjust data algorithms used to correct for potential biases or by ensuring any algorithms used by public sector employees is done in a purely advisory manner, may help dilute the negative impact of biased algorithms.
In terms of potential ways to mitigate the negative effects of automation, a few actions can be taken. One potential solution would be the promotion of skills-based retraining. As with the Industrial Revolution, increased automation will likely create more jobs than it destroys, albeit in different fields and sectors. The fields of computer programming, remote operators, and those in caring professions such as nurses or home-assistants have been growing in recent years and will likely continue to do so even as automation increases. The number of nursing assistants increased by 909%, teaching assistants by 580% and care workers by 168%, demonstrating a shift to an economy based more largely on non-routine, high-skill tasks instead of routine or non-routine low skill tasks.  The promotion of such fields, similar to the current push to encourage children and teenagers to pursue STEM fields, through early childhood programs and the integration of information about these areas into middle and high school curriculums can help prepare students for the changing future economy.
For those currently in the labor force, other options exist. Given that those in high-skill jobs likely already have college degrees or other certifications, programs such as Massively Open Online Courses (MOOCs), especially those offering verifiable certificates, offer a cost-effective way for the government to subsidize and promote job retraining. MOOCs would allow those transitioning from a routine high-skill job to a non-routine high-skill job to complete the necessary courses to be prepared for their new career without needing to complete an entire new degree and can also be completed online and while employed. Other options, such as the provision of a universal basic income, are also potential solutions for the issue of technological disruption as it would ensure consumption smoothing during transitional periods. Countries such as Finland and the Netherlands are set to begin experimenting with limited forms of basic income beginning next year and individual startups have begun pilot programs in the United States.  No matter the exact form policy solutions take, any response to the threat of automation on jobs must be responded to with an awareness that more jobs will likely be created than are destroyed and that education, including elementary and high school programs as well as retraining efforts, will likely provide the best solution to problems caused by technological disruption.
Student Blog Disclaimer
The views expressed on the Student Blog are the author’s opinions and don’t necessarily represent the Penn Wharton Public Policy Initiative’s strategies, recommendations, or opinions.
Additional Blog Posts
 A Future That Works,” Bloomberg Government, accessed August 7, 2017, file:///C:/Users/ws-jdewar/Downloads/MGI-A-future-that-works-Executive-summary.pdfhttp://www.dyogram.com/2017/04/the-jobs-will-change-and-so-will-we/.
 OECD, “The Risk of Automation for Jobs in OECD Countries,” OECD Library, accessed August 7, 2017, http://www.oecd-ilibrary.org/docserver/download/5jlz9h56dvq7-en.pdf?expires=1499438280&id=id&accname=guest&checksum=BA609B4C2A7948B645FAA85B95CC7636.
 “Re-Educating Rita,” The Economist, accessed August 7, 2017, https://www.economist.com/news/special-report/21700760-artificial-intelligence-will-have-implications-policymakers-education-welfare-and.
 Williams-Grut, Oscar. “Robots will steal your job: How AI could increase unemployment and inequality.” Business Insider. February 15, 2016. Accessed August 18, 2017. http://www.businessinsider.com/robots-will-steal-your-job-citi-ai-increase-unemployment-inequality-2016-2.
 Muoio, Danielle. “Robots could take over 80% of jobs in one country.” Business Insider. March 21, 2016. Accessed August 20, 2017. http://www.businessinsider.com/world-bank-report-outlines-countries-at-risk-for-automation-2016-3.