page image

CHAPTER II - AI, Robotics, and the Future of Work

There are things that computers are very good at, so good that they are known as “machine work.” This is structured, well-defined work that computers can tackle repetitively, without human fallibility. From weaving to number crunching, there are tasks at which humans simply cannot compete with mechanical and computational solutions. Algorithms can be applied to massive data sets to come up with desired outputs, and as more and better data points are absorbed into the calculations, more exact outputs are created. Programs that take advantage of machine learning (often conflated with Artificial Intelligence, or AI) are able to refine their own algorithms and discover patterns in data caches too large for any human to comprehend. Humans have ceded this kind of work to machines, hanging on only in the margins where no one has yet bothered to write an appropriate algorithm. Indeed, some jobs that were once thought to require long human experience—think concierge services, drafting artists, credit authorizers—have been shown to be better performed by cannily written algorithms.

Humans, however, still have their work domain. These are activities that require cognitive, social and emotional flexibility and agility, most obviously the interpersonal work that requires empathy, understanding, persuasion and care-giving. These activities are also creative endeavors, didactic and pedagogical pursuits. They are also jobs that require a degree of metacognition—the idea that humans may reflect on their own thinking process, which requires a model of consciousness that machines have not yet achieved. For computers to engage in this type of work would require true (general) artificial intelligence, something that researchers have yet to create. Human work is safe, for now.

However, there is a large category of work that can be called augmented work. At its most basic, this refers to the pairing of a human and a machine to complete a task. Maryam Alavi, Dean of the Scheller College of Business at Georgia Institute of Technology, describes this as teaming machines and human beings to do work with the goal of elevating performance to accomplish tasks that would be hard, or not possible to do by machine or human alone (see Figure 1). A category of work humans once did can now be done more accurately, more quickly and at massive scale by machines. In augmented work, an algorithm may be able to analyze patterns in big data, but it still takes human cognition to make sense of the analysis, to place it in context and to communicate the results in a way that other humans can understand and act upon it.

Figure 1. Future of Work
Source: Maryam Alavi, Georgia Institute of Technology, August 2018.

Humans are unlikely to take back many tasks that have been automated. The goal, then, is to design work and develop a human workforce that can partner with machines to perform complex and unstructured tasks. To achieve this will require humans that are educated in a particular way.

Educating the Workforce of the Future
Most educators look at learning as a mental process that is incremental. It takes place over time as a sequential process of knowledge acquisition, from basic to advanced knowledge (see the revised Bloom taxonomy displayed in Figure 2). The future of work will require enhanced levels of higher order thinking capabilities on the part of humans; however, lower level basic learning must come first. Some argument may be made that “street smarts” escape this incremental necessity. However, it may be argued that this is a form of high-level thinking built on basic lessons provided by asphalt rather than algebra.

Learning basics begin with exposure to information with proper perception and attention paid to it. Then, comes the ability to recall facts. This represents the lowest level of learning. One step up comes interpretation, or the creation of mental models based on information received. Next, is application—or the ability to use and implement acquired knowledge in new situations. These three steps constitute lower-level learning, and for many years these were the three skills that society expected educated humans to have. The modern (U.S.) K-12 educational system—which features sets of problems to be solved in pursuit of predefined answers—is largely designed to achieve competency in these areas (and to teach children how to follow directions, a skill of limited usefulness beyond a modest skill level). Many students never advance beyond the three most basic competencies.

At the next step is the level of complex analysis (i.e., analysis in various and different contexts), which includes competencies that are not easily replicable by machines. The next level calls for skills in evaluation that require the application of judgement and lead to the ability to find solutions for ill-defined problems. The next and highest step is creativity, which allows a human to realize new ideas and concepts, new connections, and new models and techniques.

In short, to create a workforce that cannot be replaced by machines—and, more ambitiously, can be partnered with machines to find and solve highly complex problems—the workforce must be competent and comfortable with higher-level learning and thinking. This highest level, the multiplier of all the others, is cognitive flexibility. In Dr. Alavi’s words, “Cognitive flexibility is the ability to do multiple things and [be] able to learn fast.” The question is, how to cultivate some competency in these higher levels of learning across diverse levels of the workforce. Everyone in the workforce will not be solving massive problems, but increasingly they will be called upon to yield machine work to the machines and use higher-level competencies to add value, think and move quickly among different concepts and contexts and [be] able to learn fast.”

Figure 2. Revised Bloom Taxonomy of Learning Levels
Source: Anderson, L.W. & Krathwohl, D.R. (Eds.) (2001). A taxonomy for Learning, teaching, and assessing: A revision of Bloom’s taxonomy of educational objectives. New York: Addison Wesley Longman.

The University Model
Traditionally, corporations have depended on universities to provide workers that are competent in higher-level thinking. The approaches provided by colleges tend to be disciplinary, with students following one path in the study of a particular subject matter (e.g. history or architecture). More recently—in the service of developing a creative skills hierarchy—schools are seeing an increase in multidisciplinary, interdisciplinary and transdisciplinary approaches to learning and education. A multidisciplinary approach involves studying and learning more than one subject matter. Examples include dual degrees in business and engineering, or business and law. In this model, a student stays in school longer and fulfills the requirements of more than one program and typically receives more than one degree. The idea is to develop capabilities to analyze and solve complex problems more effectively by applying more than one disciplinary viewpoint, opening up new courses of action and new creative insight.

Interdisciplinary learning combines the relevant and interrelated aspects of different disciplines in a single program of study. Georgia Tech has developed an interdisciplinary Master of Science (MS) in analytics degree. This degree brings together various analytical concepts, tools and techniques (from engineering and computer science) and core management courses to develop capabilities for data- and evidence-based business decision-making and problem-solving.

There is yet another approach, one that may be truly groundbreaking, as it challenges traditional pedagogy. It is known as “transdisciplinary learning,” and it echoes the realities of the post-educational world. The transdisciplinary approach is formed around a real world problem rather than core concepts or methods of a given discipline. This is inquiry-based learning, and it begins with defining the problem and searching for relevant knowledge. From there, learners back into disciplines that can provide the know-how and capabilities to solve the problem. It is clear that these are the skills that modern organizations want in their higher-level employees, where the aim is to discover something that has yet to emerge, where structures do not yet exist. For example, sustainability problems require effective and comprehensive solutions which call for transdisciplinary knowledge of social, economic, environmental and engineering aspects.

Peter Fasolo, Executive Vice President and Chief Human Resources officer at Johnson & Johnson, notes that these intersections may happen outside of the regular degree programs and curricula, which tend to be fixed. In traditional university programs, students accumulate skills in service of a degree that will convey their ability to figure out problems to be faced later. This leaves transdisciplinary learning to the domain of non-degree executive development programs. Thus, corporations can explore opportunities like, “I am in the process of X, and I need to digitize this part of the organization,” and then bring in the disciplinary perspective and new knowledge needed from wherever it resides. Teaching in these types of programs can be done by faculty from different colleges and executives from many companies.

This highest realm of learning, currently, is one that exists for the benefit of existing corporations, mostly within an “executive education program.” If the goal is to educate an entire workforce—to truly transform the way we work, and think about work—the focus cannot only be on the top tier of workers. True transformation will require a new and more inclusive culture of learning—one called “continuous learning.”

A New Culture of Learning
The half-life of skills is growing ever shorter, so learning has to be a continuous project. “Continuous learning” requires higher-level thinking skills, applied by learners in a transdisciplinary way. Learners will need to be cognitively flexible and learn fast, and in order to learn fast, they will need to know a lot already. But that does not mean just executive-level employees qualify—a whole host of workers have deep knowledge within their fields of expertise, as well as the drive to learn more. Among the skills that lifetime learning will require is the ability to recognize a solution in one domain and apply it to another—a skill that can be developed. Equally important is the concept of “unlearning”—the ability to selectively forget the tried-and-true in the service of finding a fresh entry point into a problem.

Jeff Schwartz, Principal of Deloitte Consulting (with inspiration from John Seely Brown), has suggested a model similar to DevOps, the combination of development, information technology operations and testing within the same unit. “In the same way that in technology we’ve said we’re going to take design, development, testing and operations and integrate it into a DevOps model, can we make a model in which learning is ‘Dev’ and ‘Ops’ is work?”

Workers will never have enough time to leave work, come to school, learn a few things, and then go back to work, so the system must be one in which working and learning will be increasingly intertwined—consistent, rather than episodic. Learning will become a by-product of doing the work. This will require the redesign of the work itself, and ultimately, likely the organizations and institutions where work is done.

Inclusiveness and Continuous Learning
One challenge to the continuous learning model is that it leaves a big slice of the workforce out in the cold. Maureen Conway, Executive Director of the Economic Opportunities Program at the Aspen Institute, notes that one in four working adults who work full-time do not make enough to lift a small family out of poverty. These people are adults, with adult responsibilities and adult expenses. It is difficult to get them to a level of economic stability that will give them time to learn, particularly if learning is a separate endeavor from the work. If work and learning are intertwined, this becomes less of a problem. Time, however, is not the only issue. The psychological and mental “space” to be open to learning and able to learn is important as well. Being under duress, stressed and threatened, is prohibitive of continuous learning, further widening the income and opportunity gaps.

Turning to populations that are entering or looking to re-enter the workforce, studies find that there are systemic challenges beyond just the lack of time for learning. Most notably, modern education in the U.S. is financed through loans. Because many of those loans are offered based on credit history, tens of millions of Americans cannot get access to them in the first place. Those that are able to take out loans to finance learning often find that this debt becomes debilitating upon graduation and paying it off requires them to work on a continual basis—at the expense of lifetime learning.

Moreover, incentives within higher education are rarely to students’ advantage. Colleges rely on a “butts-in-seats” business model, rewarded for putting students in classes, not for the things those students eventually achieve. Furthermore, universities—and indeed, most of society— actively promote the idea that a four-year degree is a goal worth striving for, and that this type of education should be available to everyone. Simultaneously, selectivity in admissions at elite colleges correlates with high rankings in most surveys, explicitly rewarding exclusion. All of this sends confounding messages of “everyone can go to college,” followed by “maybe not a good one,” and “probably not at a price you can afford.” Employers perpetuate these ideas by writing job descriptions that require degrees when an academic skill set may not actually be required to succeed in the job.

Starting college is a risk with few safety nets, and even successful students may struggle with debt and unemployment after they receive their degrees. But a more efficient approach—one in which students dip into the university system to gain specific skills but not a degree— saddles them with debt with none of the social and economic imprimatur that a degree conveys. Obviously, the economic reality of higher education suppresses participation by those from lower socioeconomic strata, depriving our economy of a vast pool of talent.

Until our system of higher education corrects these inequities, it is critical to develop other paths to higher-level learning. The classic model of higher education may not be the best option for many—perhaps most—people. Instead, a truly holistic system would incorporate university degrees and certifications, as well as apprenticeships and guild models. Beyond these models for education (and the augmenting of competencies as required skillsets change), workers should have access to opportunities for development through corporations which make a commitment to continuous, on the job learning.

Perhaps because of the place college holds in the popular imagination, the idea of apprenticeships gets short shrift. Colorado Governor John Hickenlooper spoke about the apprenticeship program for high school students he has championed in the state. “It’s a question of branding, and how we talk about it. One way we do not talk about it is as a substitute for college. It’s a pathway that allows you a choice of going to college or not going to college. You get to choose.”

The state’s program strives for inclusiveness, and its messaging reaches kids who are “a little different, who are willing to go out and do something like become an apprentice.” But after the initial contact is made, there is follow-up to show students—and their parents—the advantages. “We put a chart up and show, all right, if you go to Colorado State University, you’re going to spend $25,000 a year… and here’s how much you’re in the red. If you do an apprenticeship, here’s how much you’re in the black. At the end—and we do this over the course of ten years for two imaginary students on a parallel path—it’s a big delta. It’s $175,000 or $200,000 difference. And you still have the same choices.”

Employment Projections for 2016-26 from the Bureau of Labor Statistics note that, “Of the 30 fastest growing detailed occupations, 18 typically require some level of postsecondary education for entry.” Viewed from the other side of the coin, 12 of those 30 do not require college-level coursework, much less a degree. But, they will require skills. As Governor Hickenlooper notes, “The only true poverty campaign, I think, is education and training, and we’ve been saying too much education, maybe not enough training.”

Title Goes Here
Close [X]