After decades of promise and hype, artificial intelligence is finally becoming one of the most important technologies of our era. AI technologies, like machine learning, are clearly having a major impact on the very nature of work. But, how can we best quantify their impact on the actual evolution of jobs?
The best approach for exploring the relationship between technology and jobs is to look at the individual tasks that comprise a job. Most jobs involve a number of tasks. Some of these tasks are relatively routine and based on well-understood rules, while others require judgement, social skills and other human capabilities. The more routine and rules-based the task, the more amenable it is to automation.
However just because some of its component tasks have been automated, does not imply that the whole job has disappeared. To the contrary, automating the more routine parts of a job will often increase the productivity and quality of workers by complementing their skills with technologies and tools, thus enabling them to focus on those aspect of the job that most need their attention. While automation substitutes for labor, automation also complements labor, increasing productivity and other economic outputs in ways that often raise the demand for and earnings of workers.
“The emergence of artificial intelligence (AI) and machine learning (ML) poses a new set of opportunities - and challenges - for work and workers,” said The Future of Work: How New Technologies Are Transforming Tasks, a research report released last Fall by the MIT-IBM Watson AI Lab. “The tasks that can be done by machine learning are much broader in scope than previous generations of technology have made possible. The expanded scope will change the value employers place on tasks, and the types of skills most in demand.”
To shed light on the impact of AI over the past decade, the report analyzed 170 million job postings in the US between 2010 and 2017. The job postings, which came from Burning Glass Technologies, provide a detailed view of the tasks and skills employer’s were looking for. Researchers at the MIT-IBM AI Lab used natural language processing to transform the jobs postings, originally in English, into structured data sets, as well as to identify the tasks to be performed and the required skills, based on Burning Glass’ taxonomy of over 17,000 unique skills. The data set generated by these methods provides much more detail about the changes in tasks within jobs and in skill requirements than traditional survey data.
The report’s overriding finding is that while just about all jobs will change as new technologies transform their component tasks, few jobs will actually disappear. However, the way work gets done is fundamentally changing in three key ways. Let me summarize each of these changes.
Tasks are Shifting Between People and Machines - But the Change has been Small
The report analyzed the shift in the tasks employers were looking for in their job listings between 2010 and 2017, using O*NET, the most comprehensive data on US occupations sponsored by the US Department of Labor. O*NET includes highly detailed information on over 18,000 tasks that are part of 964 different occupations. The analysis found that tasks that are more suitable for automation by machine learning or other technologies are disappearing from posted job requirements more often than than those that are less likely to be automated, but the shifts are small. “Across more than 18,500 tasks, for each occupation, on average, workers were asked to perform 3.7 fewer tasks in 2017 than seven years earlier.”
The data show differences in the changes across different kinds of tasks. Job listings requested 4.3 fewer tasks between 2010 and 2017 for tasks that are more suitable for machine learning automation, - such as routine administrative tasks like scheduling and validating credentials. But the changes were smaller, 2.9 fewer tasks, for tasks that are less suitable for machine learning, - such as design and industry knowledge. While small in absolute number, this represents a close to 50% decline in the demand for tasks that are more likely to be suitable for machine learning compared to those that are less likely.
“The decreased task requirements are likely due to employers’ seeking greater focus from workers and the early adoption of AI and machine learning, indicating a fundamental shift in the way work gets done. But the shift has been small, allowing time for workers and employers to adapt.”
Tasks Increasing in Value Tend to Require “Soft Skills”
As technology automates some tasks within occupations, the value of the remaining tasks that continue to require human skills will rise, because the increased use of technology in the job will result in increased productivity and demand for worker skills.
The report found that industry knowledge tasks are on the rise in high-wage business and finance occupations, increasing in value between 2010 and 2017 by almost $7,000 per year on average, while the annual wages for manufacturing and production tasks have decreased in value by over $5,000 per year. Design-oriented tasks, - which are hard to automate because they require innovative thinking, deep experience and good judgement, - have increased in value for all wage groups. For lower-wage personal care occupations, - e.g., hairstylists, fitness trainers, recreational workers, - annual wages for design tasks have increase by $12,000 between 2010 and 2017. For mid-wage sales occupations, they’ve increased by over $8,500 over the same time period.
In a 2015 article, USC professor Ernest Wilson wrote about his research to better understand the key competencies companies are looking for. His research team gathered data from almost 2,000 executives from a broad range of industries and geographies, and discovered that companies are looking for a new kind of talent that is currently undersupplied in the workforce.
Future leaders must be strong in quantitative, technical and business skills. But these must be complemented with a unique set of attitudes, perspectives, experiences and other so called softer skills, e.g., adaptability, cultural competence, holistic thinking, intellectual curiosity, and empathy. Good leaders need to be good strategic thinkers and must have strong social and communications skills. Finding and retaining talented individuals with these capabilities is a challenge regardless of geography or industry.
High- and Low-Wage Jobs are Gaining Tasks and Earning More
The last major fundamental change in the way work gets done builds on the 2010 seminal paper by MIT economist David Autor on the polarization of job opportunities. Autor examined the changing dynamics of the US labor market by looking at three different segments: high skill, high wage jobs, where opportunities have significantly expanded; low skill, low wage jobs, which have also been expanding, while their wage growth has been relatively flat; and mid skill, mid wage jobs which have been declining along with their wage growth.
The MIT-IBM AI report found that workers in the middle segment continue to be squeezed, with tasks shifting out of mid-wage jobs into low- and high-wage jobs. “For every five tasks shifted out of mid-wage jobs, four tasks move to low-wage jobs and one moved to a high-wage job. As a result, wages are rising faster in the low- and high-wage tiers, than in the mid-wage tier. Low-wage workers gained an average of $600 in annual compensation more than mid-wage workers, while high-wage workers gained an average of $1,200 in annual compensation more than mid-wage workers over the same period.
All Jobs Will Change, says the report in conclusion. “New technologies like AI have just begun to transform work and while the rate and pace of change is slow now, it will likely accelerate as more AI solutions are adopted throughout the economy. Workers have time to adapt by learning or honing skills that require innovation, creative thinking, or deep insight and experience. Meanwhile, employers across all industries should begin to focus on reskilling their workforces, redesigning job roles and supporting career advancement.”
Comments