Q & A
A Truck Alliance Inc. office in Chengdu, China. Photo: Qilai Shen/Bloomberg via Getty Images.

What’s the Future of Work?

Mix smart machines, businesses as platforms, and diverse teams solving complex problems, add a whole lot of uncertainty, and you have a recipe for the future of work. Jeff Schwartz ’87, a principal at Deloitte, discusses how leaders can navigate fast-approaching opportunities and challenges.


The robots are coming! A Pew Research Center survey found that 72% of Americans are concerned about robots and computers taking jobs currently done by humans—though just 2% report having actually lost a job to automation.

In a conversation with Yale Insights last year, Yale SOM labor economist Lisa Kahn said that the effects of automation are already being felt throughout the economy. In fact, the Great Recession likely accelerated the process, both by slowing demand, giving firms a chance to retool without losing sales, and by providing an excuse to lay off unproductive workers.

“We have seen the influences of automation in almost every pocket of the labor market, from the very low end of the skill spectrum to the very high end,” she said. “The future of labor essentially comes down to, where are computers going to replace us and where are computers going to augment us?”

The New Yorker said that economists have long believed that technological changes eliminate some jobs but create plenty of new ones to replace them. Now they aren’t so sure. MIT’s David Autor has found that automation fundamentally alters the supply-and-demand equilibrium. “A subset of people with low skill levels may not be able to earn a reasonable standard of living based on their labor,” he told the magazine. “We see that already.”

So what does it take to keep a job? Kahn’s research has shown that cognitive and social skills are key: “If you have one of them, and especially both of them, I think it’s very likely that you’re going to be pretty safe from automation for a long time.”

To find out more about the future of the work, and how artificial and human intelligence can co-exist in the office, Yale Insights talked with Jeff Schwartz ’87, a principal for Deloitte Consulting and the company’s global lead for human capital marketing, eminence, and brand.

Q: What are the key forces shaping the future of work?

Two megatrends are driving the future of work. One is that organizations are being dramatically reoriented and restructured. The historical view of an organization as a hierarchy is being replaced by a view of the organization as a network or an ecosystem. Instead of divisions, functions, or processes, organizations are increasingly being built around teams.

The second big shift has to do with work itself changing. An increasing number of tasks are being accomplished through automation or cognitive computing. To simplify it greatly, if we can articulate the process of something, we can automate it. It’s my expectation that in the next five to ten years everyone will be working next to and with a smart machine they’re not working with today.

Q: What is the role of people in this emerging future?

The question that we’re looking at in every company and every industry is, what are the essential and enduring human skills? What are the things that smart machines can’t do? I’m not quoting it correctly, but Pablo Picasso said something like, “Calculating machines are useless. They can only give you answers.” Asking questions is an essential human skill. I’m not talking about the kind of questions a chat bot can manage. I’m talking about the sorts of creative thinking and inquiry that lets us frame a problem.

For a range of reasons, the problems facing businesses and the public sector are much more complex and multi-disciplinary. The complexity of the problems and the pace of change means we need to work collaboratively, on teams. Working in teams is itself an essential human skill.

The relationship that we're developing with smart machines is different than before. We’re getting a glimpse beyond the digital native to the AI native who doesn't think twice about talking to their phone or any other device. We're getting to the point where natural language interaction—talking to our machines, our machines talking to us—is rebalancing work roles and the ways machines and people can augment each other.

Q: What does it look like to team with smart machines?

Let me offer two examples. Both have been highly popularized, so they should be in some sense familiar. One is IBM’s Watson platform as it relates to medicine. The way the IBM team put it is that after Watson won Jeopardy they sent it to medical school. That meant they fed Watson medical journal articles and data. They developed its capability to read radiology reports and to do oncology diagnoses. At this point, in terms of diagnosis accuracy, the average doctor is around the 50th percentile while Watson is 75th percentile.

Additionally, there’s an explosion of knowledge and data. A really great physician can read a couple hundred journal articles a year, but in any particular field there are thousands of articles written. Teaming with a machine learning technology like Watson could bring all that technical information to bear in diagnosis and coming up with potential treatments while doctors decide on the most appropriate approach and explain to a patient what a diagnosis means, something that currently is very difficult for a machine.

A very different industry, financial services, offers another example. My colleagues Tom Davenport and Jim Guszcza wrote about this recently in the Deloitte Review. One of the examples they use is robo-advisors. They are algorithms, basically, to help create financial portfolios.

Some investors interact directly with robo-advisors online. But many consumers are more comfortable interacting with a person. The financial advisor can still leverage the algorithm’s ability to do the calculations involved in setting up portfolios that meet the desired criteria while focusing on how she or he relates to the customer and ultimately working with more customers.

What these two examples have in common is that there are parts of knowledge work—data sorting, pattern matching, or algorithmic calculations—suited to the machines on our team. Other things are suited to the people on our team. The idea is to work together to augment each other.

Q: How are organizations changing?

What’s an organization going to look like in the 21st century? I think it’s up for grabs. Ronald Coase won his Nobel Prize for telling us in the late 1930s that firms were largely based on transaction costs. Our ability to transact and interact on internet-based platforms has blown away some of our concepts about transaction costs. Work and jobs are being separated from companies because there’s something competing with the traditional corporate organizational form, which is platforms. Beyond thinking about the key design principles of teams, networks, and ecosystems, we need to explore what it means to be a platform-based organization and what it means to be an asset-light organization.

It’s a taxi cab company taking out ads in the Yellow Pages, hiring dispatchers and drivers, and maintaining a fleet of taxis versus Uber, which owns no cars and doesn’t have any drivers. All they have is a platform that connects people that need rides and people that want to provide rides.

The boundaries of organizations in the 21st century are going to be shaped by companies that are the intermediaries between producers and consumers. Or, for many of us today, the platforms that let us constantly shift back and forth between being producers and consumers.

Q: So much of the discussion seems to split the future into either robopocalypse or techno-utopia? Is it that much of a binary?

In 1930, John Maynard Keynes wrote “Economic Possibilities for our Grandchildren,” an essay on the future of work. He envisioned that within 100 years the average person would be working 15 hours a week, and we’d have an excessive leisure problem. Obviously, that’s not what happened.

Keynes was a brilliant economist, but he missed something. He missed the number of new fields and human endeavors that would be invented—modern health care, education, and the technology industries, for a start. That’s largely what has driven employment and progress. That’s one of the things that make many of us who are looking at the future of work optimistic. The future doesn’t have to be about grinding out efficiency; it can be about exponential innovation.

It’s less a question of whether robots take our jobs, and more a question of how robots and technology will change our jobs. Automation, and not just robotics, but cognitive technology and AI, natural language processing, and machine learning will take some jobs while also creating new ways of working and extended labor platforms. I think it’s reasonable to expect that all of our jobs and all of our careers will be significantly or fundamentally changed by technology and different labor options.

As that is happening, we have the opportunity to redesign organizations, work, and how we think about careers and learning. If you’re motivated by the idea of living in fast-changing times, it’s going to be pretty exciting. If you’re in fear of fast-changing times, it’s going to be tough.

Q: Today’s cars are huge improvements over Ford’s Model T, but they aren’t so different that we can’t trace the lineage. To what degree have we seen where today’s technology and innovations will take us?

The short answer is, I don’t know. But I agree with William Gibson, who says the future is here, it’s just unevenly distributed. I’m not sure we know what today’s equivalent of the Model T is yet. If we could identify the Model T, the thing that we will incrementally improve far into the future, then we could extrapolate out.

I do think we have an idea of what the 21st century’s drivers might be. Mass production was the driver that enabled the Model T. Platforms, and the way that we interact in the economy, may be one primary driver for the 21st century.

In the early 1960s, Gordon Moore postulated basically that computing power doubles every 18 to 24 months. We’re now 25 to 30 turns into Moore’s Law. When you move something exponentially 25 or 30 turns, every additional doubling is a massive increase in computing and processing power. It makes possible things that simply were not imaginable, economically or technologically, before. We’re applying that to so many domains right now, it’s hard to know where that will go, but exponential technology is a likely another driver.

A third potential driver is in some ways the opposite of the Model T: extreme customization through technologies like additive manufacturing, where the setup costs for creating a different item is practically zero, so that we can both mass customize and mass produce. I think the 21st century will be in some way driven by platforms, exponential technology, and mass customization.

But the fourth driver, which is probably the biggest, is uncertainty and surprise. I was struck by the Queen of England asking after the financial crisis how all the brilliant economists in the UK and around the world, how the profession as a whole, missed something as big as the financial crisis. I think there’s some element of this now in every field. The interaction of highly complex global systems with the uncertainty and unpredictability of human behavior on a massive scale creates the potential for surprise that can happen extremely quickly and be very pervasive. I think that’s only going to continue. It may be fascinating or terrifying—probably a little bit of both.

I don’t think we have yet seen the potential and positive disruption of exponential technologies and platforms play out. I think the next 20 or 30 years will be about new technologies and platforms coming online, but also about adoption and pervasiveness as these ideas work their way into the economy.

Q: What does this mean for individuals?

Learning is the job in the 21st century. Full stop. As Lynda Gratton and Andrew Scott at London Business School tell us in The 100-Year Life: Living and Working in an Age of Longevity, right now, we’re living on average into our 80s, but millennials can expect to live to 100. What does it look like when the average time in a job is four and a half years and the half-life of a learned domain skill, like a computer language, is five years? How many careers are you going to have?

Adaptability and continual learning are not among the core skills; they are the core skills. Learning new domain knowledge is important, but we can learn it relatively quickly and we can access it highly effectively by collaborating with smart machines. More importantly we need to develop essential human skills, and we need the skills and capacities that let us partner in a team with machines. Individuals and companies need to be organizing around learning. And the learning needs to be organized around dynamism and change at speed.

Individuals have a responsibility for maintaining and developing their own skills and retooling through lifelong learning. Businesses will be benefited by being thoughtful about how they redesign jobs and teams and find ways to facilitate learning. But there is a significant set of responsibilities for government, public policy, and social institutions.

Q: What should that look like?

I recently co-wrote an article titled “Navigating the Future of Work” with John Hagel and Josh Bersin. Our observation is that if work is being augmented by on- and off-balance-sheet workers along with machines, in order to benefit from the incredible exponential power of technology and get our arms around the challenges that come along with it, we need to consider how we support people in these different arrangements.

Part of that is helping people through economic transitions. That includes healthcare and different kinds of income insurance. We are going to need really good data on people in a gig economy, so we need to improve the data that we gather on employment, education, and skills.

We need to recognize that each of us will need to do some version of going back to school every decade. We need to ask fundamental questions about how communities, cities, states, and the federal government support that. We need to look at every segment of the population and ask the question: what types of tuition credits, tax credits, or new forms of community college would incentivize people to educate themselves? That includes people in their 50s, 60s, and increasingly people in their 70s and beyond.

Q: Nearly every era thinks the challenges it faces are different. Is this time different?

We all decide every day, as students, employees, as business and government leaders: are we seeing incremental or exponential and transformative change? Are you going to bet on marginal changes during the life of your career, or are you going to bet on the world changing quite significantly? Will all the jobs be taken by machines or will more new jobs be created in your life than we’ve seen in the last couple of thousand years? We’re all dealing with those questions every day.

I don’t have the answers, but I do have two daughters who are 23 and 25. I’m hopeful about what might be in front of them and what will be in front of our grandchildren. It’s going to be a wild ride. The things that we’re talking about, exponential technology, platforms, uncertainty driven by the interconnectedness of the global economy and global systems can make wonderful things happen, and there will also be real difficulties.

Principal, Deloitte Consulting