On October 31, I participated in a symposium sponsored by MIT’s Center for Digital Business on The Race Against the Machine: The Future of Technology and Employment. The symposium was associated with an event that took place later that same day on the Harvard campus, the IBM Watson Challenge. The Watson Challenge was an exhibition Jeopardy! match between Watson, IBM’s question-answering system, three students from the Harvard Business School and three students from MIT’s Sloan School of Management. Watson won the exhibition match, with the Harvard team finishing a close second.
At the MIT symposium, I was part of a panel on How will technology affect productivity and employment? The panel was moderated by MIT management professor Erik Brynjolfsson, who is also the director of the Center for Digital Business. The other panel members were MIT economists David Autor and Frank Levy. We discussed the impact of technology on jobs, in particular, the impact of advanced systems like Watson which are pushing the boundaries between human and machine capabilities
Will there be enough good jobs to go around? “The coming world war is an all-out global war for good jobs,” argues Gallup’s Chairman and CEO Jim Clifton in a provocative new book, The Coming Jobs War. In this recent Forbes interview he said:
“Gallup has discovered that having a good job is now the great global dream; it’s the number one social value for everyone. This is one of our most powerful findings ever. A good job is more important than having a family, more compelling than democracy and freedom, religion, peace and so on. Those are all very important but they are now subordinate to the almighty good job. . . According to Gallup’s World Poll, there are three billion people out of seven billion who want a good job. There are only 1.2 billion jobs to go around. So there’s a short-fall of 1.8 billion jobs. The question is who gets those new jobs as they emerge.”
These are not new concerns. For example, in a paper written in 1930, economist John Maynard Keynes wrote: “We are being afflicted with a new disease of which some readers may not yet have heard the name, but of which they will hear a great deal in the years to come - namely, technological unemployment. This means unemployment due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour.”
In past technology-based economic revolutions, the periods of creative destruction and high unemployment eventually worked themselves out. Over time, the same disruptive innovations responsible for the technological unemployment disease led to the transformation of the economy and the creation of new industries and new jobs. While we are hopeful that this will once more be the case, there is no way of knowing. Perhaps, this time around, the dramatic advances in technology, coupled with the forces of globalization will make it very difficult for jobs to recover in the US and other advanced economies.
To explore the impact of Watson-like technologies on the labor market, we need a framework to help us categorize the different kinds of human work skills. This will help us analyze how advances in technology will impact work by substituting for or complementing human skills over time.
I like the framework proposed by my fellow panelists Frank Levy and David Autor in two different papers: The Skill Content of Recent Technological Change, published in 2003 by Autor and Levy along with Harvard Professor Richard Murnane; and How Computerized Work and Globalization Shape Human Skill Demands published in 2005-2006 by Levy and Murnane.
Human work tasks can be viewed along two dimensions: whether they are cognitive or manual; and whether they are routine or nonroutine, resulting in a 2 X 2 matrix:
Routine, cognitive tasks are those human information-based activities that can be well described by a set of rules. They include performing calculations, record keeping, dealing with simple customer service questions, and many kinds of administrative tasks. These white-collar activities have been prime candidates for technology substitution or automation, as well as for offshoring to lower-cost countries.
Routine, manual tasks are those physical tasks that can also be well described by a set of rules. Many manufacturing activities in assembly plans fall into this category, as do packaging pills into containers. These blue-collar activities have also been prime candidates for technology substitution and offshoring.
Non-routine, manual tasks are physical tasks that cannot be well described by a set of rules that a machine can follow. Many low skill, low pay activities fall into this category, such as janitorial services, gardening, fast-food restaurant jobs and health care aides. A number of better paid jobs also fall into this category, including driving a truck or taxi, and building jewelry. These activities are not candidates for technology substitutions. They are also not so easy to complement with technology-based tools.
Non-routine, cognitive tasks comprise the final category. These are generally high-skill human activities that involve expert problem solving and complex communications for which there are no rule-based solutions. Examples of expert problem solving include sophisticated medical diagnosis, complex designs, many R&D tasks, making decisions in complex organizations, software architects, and diagnosing tough auto repair problems. Examples of complex communications include managing and leading large groups, teaching, writing books and papers, arguing legal cases, and selling sophisticated equipment.
These non-routine cognitive tasks are beyond the scope of computer substitution for the foreseeable future. However, one can design sophisticated tools to significantly expand what people can accomplish when performing these activities. For example, CAD systems have enabled engineers to develop far more complex products than they could have done otherwise. Social networking tools make it easier for people to communicate and collaborate with colleagues over wide distances and thus improve the collective intelligence of the team. And, systems like Watson will be extensively used to help experts deal with highly complex problems in areas like medicine, finance and national security.
Routine manual and cognitive tasks have been the most impacted by technology advances, including the offshoring of jobs, which itself is the result of the rise of high bandwidth global networks. As per David Autor's 2010 paper, The Polarization of Job Opportunities in the US Labor Market, this has resulted in declining job opportunities for both middle-skill white-collar clerical, administrative, and sales occupations; and middle-skill, blue-collar production, craft, and operative occupations.
The creation of large numbers of such mid-skill, mid-pay jobs was a major US achievements in the 20th century. It gave rise to the employment of large numbers of blue collar workers in manufacturing plants, and large numbers of white collar workers in business and government offices. But, the same well defined, standardized, industrialized processes that resulted in the creation of all these jobs, made it possible to automate many of the tasks as information technologies become more powerful and inexpensive, as well as to re-engineer the organization and either eliminate or move many of those jobs to less expensive regions around the world.
Non-routine tasks are a totally different story. Despite the dramatic advances in technology, there will likely still be plenty of such non-routine jobs once the economy begins to recover. But, to create such jobs, nations and regions need to seriously invest in a number of key areas, starting with education.
There was a consensus in our panel, as there has been in just about everything I have read in the past few years, that a significant increase in educational attainment will be required for many of the kinds of jobs that will be available over the next few decades. Another finding in David Autor's 2010 study is that over the past few decades, opportunities have continued to expand in high-skill, high wage jobs, as have the earnings of the college educated workers that fill such jobs.
Higher educational attainment is also increasingly required for many of the middle skill jobs opening up in areas like healthcare, diagnosing and repairing complex equipment, and in many trades which are now using sophisticated software-based tools. Such jobs require a post-high school education, but not necessarily a traditional four year college degree. Good community colleges and vocational schools could actually provide better training for such jobs because the skills needed are more practical and hands-on than the more abstract skills taught in many college programs.
To help prepare the labor force for the kinds of non-routine cognitive and manual jobs we expect to see more of in the future, requires a national commitment to education and job creation in general. Most (non-ideological) economists have concluded that assuming that markets can fix the job problems by themselves is not realistic because of the serious structural issues involved. If not properly dealt with, high unemployment, stagnant wages and a sharp inequality of incomes could truly unravel America’s social contract and lead to an increasingly unstable society.
In the end, any discussion of jobs in the age of the kind of dramatic technology advances exemplified by Watson, requires us to reflect on the kind of society we aspire to become, and what we are willing to do to attain our aspirations.