August 6, 2014

AI, Robotics, and the Future of Jobs

Views from Those Who Expect AI and Robotics to Displace More Jobs than They Create by 2025

On the other hand, 48% of our survey respondents agree with the statement that “networked, automated, artificial intelligence applications and robotic devices [will] have displaced more jobs than they have created by 2025.” Their responses incorporate several major themes.

Advances in technology will absolutely reduce human jobs—this process is already underway, and the logic of our economy and technological advancement make it a sure thing to continue

Many of the experts in our survey who expect technology to be a net job destroyer also looked to the history of technology and employment in making their case. But in contrast to the group discussed above, these respondents see a much different story—one in which advances in automation have been taking jobs and putting downward pressure on wage growth for years.

Alex Halavais, associate professor of social and behavioral sciences at Arizona State University, predicted, “They have probably already replaced more jobs than they have created. The slow recovery in the U.S. is closely tied to our worker productivity, which is in turn related to our use of technology. We are only at the cusp of this, and I suspect it will be far more obvious and pronounced by 2025. I suspect that ATMs and self-checkout are just the starting points. The biggest shift will be fairly invisible in the next 10 years, because they will be in manufacturing, particularly at small scale. Tesla’s factory is new standard for relatively small-scale production. As some kinds of standardized service jobs become more easily addressed by scalable technology, they will go the way of phone operators and bank tellers. That is, they will not disappear entirely, but they will be radically reduced. Right now, things like food service, travel, and hospitality are being kept human for cultural rather than purely economic reasons, I suspect.”

Future of AI/robotics

Larry Gell, founder and director of the International Agency for Economic Development (IAED), responded, “After 50+ years working for the heads of the world’s biggest corporations all over the globe—watching them cut costs every place starting with the biggest cost: PEOPLE; moving labor to cheapest markets, then replacing them as fast as possible with robots and automation—why would it stop? It will accelerate. Anything and everything that can be automated to replace humans will be done. You can bet on it!”

Mary Joyce, an Internet researcher and digital activism consultant, replied, “To the extent that human workers can be replaced by robots and algorithms, they will be. There’s no reason to believe that firms would behave in any other ways. And social forces, like unions, that would limit these actions, don’t have the strength to prevent these changes.”

Karl Fogel, a partner with Open Tech Strategies and president of QuestionCopyright.org, wrote, “The reason people are investing in machine agents is precisely that they will replace more (lower-paid) humans than the number of (more highly-paid) humans needed to build and maintain the machines. But this is not a new phenomenon—it’s been going on for more than a century. We’re going to have to come to grips with a long-term employment crisis and the fact that—strictly from an economic point of view, not a moral point of view—there are more and more ‘surplus humans.’”

John Wilbanks, chief commons officer for Sage Bionetworks, wrote, “There remain enormous market gaps where digital tools can replace people, from parking lot attendants to call centers to checkout lanes in retail. Those jobs will go and won’t come back.”

A distinguished engineer working in networking for Dell wrote, “It’s a given that computers will get more powerful and be able to perform more and more intelligent tasks. This is going to create more unemployment and I’m not sure how all this gets resolved.”

Dean Thrasher, founder of Infovark, Inc., wrote, “More and more fields seem ripe for automation, but it’s hard to think of areas of our economy that are suffering from lack of staff—possibly teaching or healthcare? Yet we are applying more robotics and AI in these fields as well. I think technology’s negative impact on employment is likely to grow worse in the near future, rather than better. It’s easier to think of the few areas that will be resistant to robotics: sports leagues, symphony orchestras, craft brewing, ballet, and fine art. If the human touch is not essential to the task, it’s fair to assume that it could be automated away.”

Bernard Glassman wrote, “I’m honestly trying to think about the last time I heard anyone of any importance argue with a straight face that we should adopt a new technology because it will create jobs. At best, new robotics technologies move people into the service sector, at least until the service itself can be automated. Take 3-D printing—can we honestly believe that it will generate more high-level jobs than it kills?”

Lyman Chapin, co-founder and principal of Interisle Consulting Group, wrote, “Anything that can be automated will be, and to a greater or lesser extent depending on circumstances businesses will be reluctant to hire people to perform tasks that can be performed by robots, digital agents, or AI applications.”

Dave Kissoondoyal, CEO for KMP Global Ltd. and Internet consultant active in Internet governance activities, wrote, “We have already witnessed the effects of mechanization and automation on the labor force. Similarly—networked, automated, artificial intelligence (AI) applications and robotic devices will have displaced more jobs than they have created by 2025. The effects this time will be for both white and blue-collar jobs.”

Mark Johnson, CTO and vice president for architecture at MCNC wrote, “The trend towards automation of every job seems inexorable. This probably has a disproportionate effect on older workers of all kinds who are less agile in their ability to move about in the economy than younger workers.”

Serge Marelli, a past member of IEEE and ACM, wrote, “Automated cars can cheaply replace public transportation drivers, and automated cleaners and caregivers might very well replace human-help and caretakers for the sick and elderly. While this may seem in the short term more economic, it will be a fatal blow to the last local jobs for those with less skills (formal education).”

These advances are different from what has come before them—the changes are more rapid, and are going to impact people and professions that have thus far been insulated from automation

Many respondents worry that the current wave of technological change is going to impact previously insulated professions, and will happen so quickly as to prevent people from adjusting to new career paths.

Jeremy Epstein, a senior computer scientist at SRI International, responded, “The net number of jobs displaced will be fairly small, but they will be disproportionately blue-collar and pink-collar jobs going away and new white-collar jobs created. Just as travel agents (a pink collar job) have been largely replaced by Kayak and the like, many other service jobs like taxi drivers will largely disappear. There are no elevator operators left in the Western world (I’ve seen them still in India, though), why would anyone need a human to pilot a car to a location? Having a human driver may be seen as a status symbol for the wealthy, but even they will see the value in not having to worry about their driver’s sobriety or willingness to share overheard secrets. Blue-collar jobs like construction will still exist, because the costs of automation are too high. However, even they will be reduced as there’s more factory-built housing, which allows for cost effective use of robots in the construction process. It’s hard for me to guess how things like garbage collecting will be affected—use of equipment has reduced the number of people involved, and self-driving vehicles could reduce it further, but given the low wages it might not be worth eliminating people altogether.”

Joel Halpern, a distinguished engineer at Ericsson, wrote, “While the advent of automated assistive technology will enable many new jobs, it will likely render irrelevant many current jobs. I expect that in the same time frame other technologies will likely create many opportunities, but in terms of the direct job destruction, creation, and disruption from automated operational technologies such as implied by the question will likely be negative in terms of numbers of jobs. While the effect will be felt more on the ‘blue-collar’ level, it will likely also occur at the ‘white-collar’ level as well.”

An attorney at a major law firm responded, “The field within which I work currently employs many thousands to review documents. They are already being replaced by predictive coding algorithms. By 2025, those jobs will not exist for any but the most opaque documents and thus there will be many thousands of lawyers out of work. I find it difficult to imagine any industry which is more knowledge and thought intensive than law and we are already being replaced by machines. I suspect this will disrupt most industries.”

David Allen, an academic and advocate engaged with the development of global Internet governance, replied, “The underlying, fundamental determinant is rate of change, between invention and the workforce. The last century plus has seen the most phenomenal acceleration in the rate of change for innovation. The rate seems likely to continue high. On the other hand, people change and adapt to these changes in the real world only with difficulty. If this is correct, then the rate of change in invention will continue to overwhelm the ability of people—in this case the workforce—to adjust to that change.”

The CEO of a company that makes intelligent machines to make you smarter about your money wrote, “Most information work isn’t all that complicated. Rarely, in fact, does it require the kind of creative manipulation of symbols that usually counts for human intelligence. Where such tasks can be automated, they will be with an appropriate reduction in the human effort required. We’re just seeing the first fruits of this automation today, in fields like banking, where traditional retail banking services have been reduced to a couple of clicks in a mobile application—who needs a branch teller when you can have that teller in your pocket? This goes double for truly mundane tasks like securities trading, where algorithms running in server farms located in the same co-lo as the exchanges execute 50% of a day’s trades on many markets. Where the money goes, so goes the society. I expect the service industries will survive for another 50 years or so past 2025, but then they will be ripe for automation as well, once we can build computers that can process natural human language more accurately, and robots that can simulate human behavior more closely.”

A university professor from the United States wrote, “The impact of AIs and robotics is often, I think, overstated, but automation of vehicles and improvements in robotics in warehousing operations should lead to a steady loss of employment in all areas of logistics, with the impact felt initially in warehouse operations and then moving into delivery of goods/materials. If Amazon is already seriously contemplating delivery-by-drone, I cannot believe they are not also planning on automating warehouse operations to a greater extent than they already have.”

Mike Osswald, vice president for experience innovation at Hanson Inc., wrote, “Many jobs—truck drivers, customer support, light assembly, bank tellers and store checkout staff—will be diminished for businesses who can afford the upfront implementation costs. People will be displaced, businesses will slowly transition their older workforce to different jobs and not hire younger people or veterans. Businesses who let go of many people when adding robots will face backlash from citizens, but only for a time.”

Tom Folkes, an Internet professional, replied, “We will shortly be able to replace low level information workers—these being teachers, lawyers and librarians. In the not distant future, taxi, bus, and truck drivers. Delivery and food workers will be replaced by 3D printing. The number of people required to develop these systems will be relatively small.”

As the split between highly skilled workers and others continues to grow, current problems with inequality are going to get even worse

A number of these experts offered thoughts on how advances in AI and robotics may lead to increased income inequality and contribute to the ongoing hollowing-out of the middle class.

Bob Briscoe, chief researcher in networking and infrastructure for British Telecom, replied, “Robotics is more likely to have displaced blue-collar jobs, deepening the divide between the haves and the have-nots, and protecting the ‘haves’ from withdrawal of labor and similar industrial action. Rather than increasing leisure time, the ‘haves’ will use the freed-up time to achieve more, because maintaining the previous level of achievement would be rewarded less (relative to a living wage). The greater intensity of economic activity will maintain employment for blue-collar workers, but with similar levels of unemployment as today.”

Robert Cannon, Internet law and policy expert, wrote, “During the Industrial Revolution, although Adam Smith will disagree, our economy has been based primarily on labor. The Industrial Revolution displaced labor from agriculture to the city—but the labor existed. Where there was work to be done, humans were the best “machines” to do the labor. The humans would be paid for their labor; the humans would then pay for goods produced by other people’s labor. As production became more efficient, labor continued but moved into non-essential vocations (where essential is food and shelter). In the future, that foundation of our economy—labor—will be gone. Humans will not be the best “machines” to get work done. What will be left? Capital (ownership) and creativity (human contribution), and perhaps competition (sports, other competitions of humans as we are keen on the realization of the best among us). This will be a massive displacement of the middle class. There will be an ownership class and there will be a poor class that works at a rate below what would economically justify bringing in automation.”

S. Craig Watkins, a professor and author based at the University of Texas-Austin, replied, “This is already happening and while the rise of intelligent machines will contribute to the loss of jobs it will also create new jobs—managing, designing, building, and managing the new systems that will emerge. The challenge is will those new jobs require high skills that only a select portion of the population will be able to acquire? In general, the jobs loss will not likely be matched by the jobs created, thus creating a net loss of jobs overall.”

Henning Schulzrinne, an Internet Hall of Famer and technology developer and professor at Columbia University observed, “Many routine information aggregation and information routing jobs (e.g., in sales, customer support, health care and legal support) will be endangered, as well as some janitorial tasks. I don’t see self-driving cars displacing livery or truck drivers, as they are more likely to be used for parts of driving (e.g., on interstates) or to support drivers. You still need to unload delivery trucks, for example. However, in some cases, jobs won’t be replaced, but rather be down-skilled or bifurcated into a small number of high-skill, high-pay and a much larger number of low-skill, low-pay positions.”

John Anderson, director of broadcast journalism at Brooklyn College, wrote, “It’s the same pattern we saw in manufacturing: the de-skilling of some forms of work due to improvements in technology. The social consequences are also the same: displacement, increased insecurity, growing inequality.”

A professor of communication at the University of Southern California and well-known researcher of Internet uses and users replied, “I worry that these technological developments will further erode opportunities for working class labor in the United States and around the world, further destabilizing the employment situation for many people and further exaggerating the divide between have and have not. I don’t think smashing the machines has ever worked as a response to such developments, but this points all the more urgently to the needs of governments and citizens to more directly address inequalities in economic opportunity.”

A private law firm partner specializing in telecommunications and Internet regulatory issues wrote, “The ability of robots and AI to take on many basic tasks and jobs will relentlessly increase. That means that our total output/production may well increase even as the number of people required to generate that production goes down. That will create vexing problems of distribution of wealth/income, as the folks who own the robots etc. will claim entitlement to all or nearly all the production—yet the ability of people to buy that production will be in the aggregate declining. Over time (again, decades, not 11 years) I suspect that there will be a move towards, and an increase in the value of, unique personal-service type jobs. But that will simply highlight the conflict between different groups.”

Future of AI/robotics

A college professor wrote, “This has already begun happening. If we’re lucky, we’ll all be put on middle-class welfare to keep people from becoming destitute and desperate. We are not creative enough to make meaningful jobs out of nothing—and that’s what we’ll be left with when we give all the skilled labor and unskilled labor to the machines.”

An Internet engineer and machine intelligence researcher responded, “With the erosion of manufacturing and manual labor jobs, the underpinning economies of the lower and middle classes have been and will continue to be undermined. Wealth will continue to migrate towards the select few who have control over information resources. The control of information will be markedly enhanced by advances in machine intelligence.”

Mikey O’Connor, one of two elected representatives to ICANN’s GNSO Council, representing the ISP and connectivity provider constituency, wrote, “There will always be a LOT of jobs that are more cheaply performed by extremely low-wage humans than technology. Life at the grinding bottom of the income ladder will be largely unchanged, with any hope of improvement coming from other sectors and technologies. Life in the middle will be changed dramatically. A decreasing few will graduate into wealth and comfort, while most will slip towards the bottom. The middle will continue to become a smaller proportion of the population. Robotic and AI technology, once hoped to mitigate this trend, again disappoints. Professionals are coming under increasing pressure and have joined the middle class on the knife-edge between jumping up or sliding down. Their lives will become ever more stressful as they fight to maintain their position. Life at the top will not change much, although it will be more luxurious (if that’s possible to imagine).”

Oscar Gandy, an emeritus professor at the Annenberg School, University of Pennsylvania, wrote, “If ‘displaced’ means or includes ‘replaced with lower paying jobs’, there is no question in my mind about that: this is a process already clearly visible. While not the only determinant, the hollowing out of the middle class that we are seeing is due in no small part to the replacement of mental/creative/analytical workers with software/systems. This can only increase.”

A retired software engineer and IETF participant responded, “To the extent that our culture focuses on monetary value, and to the extent that labor cost has become the primary dimension in which Western corporations are able to optimize, the only way that automation will be permitted to create more jobs than it destroys will be if those new jobs are at substantially lower wages than the existing ones.”

Stuart Umpleby, a systems theory expert and professor at George Washington University, sees these advances leading to a new type of digital divide: “It is very easy to make a digital device that will make a routine decision. This frees up time to do other things. However, it also makes life more complicated, because one then needs to monitor and control one’s digital agents. It also requires a different type of thinking. For example, instead of going to the store to buy food, one needs to learn how to sign on to a website, order food, monitor delivery and payment. One lives increasingly in an informational environment rather than a physical environment. A virtual environment is more easily monitored by businesses and simulated by scam artists. People must learn how to identify scams, which most likely will become more sophisticated. The gap between those who live primarily in a virtual world and those who live primarily in a physical world will grow.”

We run the risk of creating a “permanent underclass”

A notable number of respondents expressed concern that we will see the emergence of a large class of people who have lost their jobs to automation, and who have little hope of gaining the skills needed to obtain meaningful employment in the future.

Bill Woodcock, executive director for the Packet Clearing House, responded, “We’re seeing AI and expert systems beginning to replace or augment customer-service jobs now, and that trend will continue. I believe that’s a good thing, as they’re replacing jobs starting with the most tedious, leaving the ones that require the most critical thinking and ingenuity for humans. As always, people will find ways to occupy themselves, and I believe AI are not a problem here. Far more troublesome is the trend toward greater social divide, that leaves a larger portion of the world’s population in poverty and unable to garner any advantage from self-driving cars or robot vacuum cleaners, because they simply can’t afford cars or vacuum cleaners of any sort, nor services that come with customer service, whether AI or human. Implicit in this question is an assumption about a middle class that still makes up the bulk of the population of Western nations, and to which many developing countries aspire, but which is, in reality, facing a decline if current trends continue.”

A technology writer observed, “Look at yard maintenance, which employs hundreds of thousands. As soon as there’s a safe, cost-effective, lawn-mowing robot, that robot will take over all the lawn mowing jobs there are. Artificial intelligence that will be able to answer questions over the telephone will displace the average call center employee for most calls. Those with only minimal education will be forced even more to the margins of society. Likely there will have to be a new social safety net for those that are simply unable to earn more than a poverty wage.”

Mark Johns, a professor of media studies at a liberal arts college in the U.S., said, “Many manufacturing and service jobs will be eliminated by intelligent agents in the next decade. Social problems associated with a growing “underclass” will increase…The middle class will continue to shrink, and there will be a greater gap between the educated and tech-savvy ‘haves’ and the uneducated ‘have-nots’.”

The research director at a technology trade association responded, “More people will be forced out of growing sectors of the workforce, with downward mobility, unemployment and underemployment resulting. Growing alienation and fear of the future will mark the lives of some members of the baby boomer population. Traditional jobs across the board, from entry-level service jobs through higher-skilled production and intellectually-challenging jobs, will be reduced in number.”

Jamais Cascio, a writer and futurist specializing in possible futures scenario outcomes, sees this new underclass having a gender component when he writes, “Unlike the numerically-controlled factory robots of the 1970s, today’s general purpose machines are designed to be easily-adapted to new job requirements. It won’t just be dropping a robot into the human’s seat…The self-checkout system at many grocery stores is a perfect example of what I mean: we didn’t just build a robot checker, we made machines that split the checkout task with the customer. Digital travel websites replacing travel agents is another example. We’re already seeing some grey-collar and specialized white-collar jobs being absorbed by machines, from legal assistants to surgeons. I expect that to continue, even accelerate. The biggest exception will be jobs that depend upon empathy as a core capacity—schoolteacher, personal service worker, nurse. These jobs are often those traditionally performed by women. One of the bigger social questions of the mid-late 2020s will be the role of men in this world.”

Dan Coates of YPulse responded, “A great thinker in this space is Tyler Cowen who in his book Average is Over outlines a dual track economic reality wherein those who leverage automation enjoy an escalating standard of living, while those displaced by automation descend into a dramatically reduced standard of living.”

A doctoral student in information science at the Universidade Estadual Paulista, in Sao Paolo, Brazil, wrote, “Robots and automatization will only release qualified personnel from heavy duties. But big masses of unqualified people will still be available, but now, competing with machines. Wages will reduce as well as labor protection in advanced countries. In undeveloped countries the situation will remain the similar as today. Big masses of poor people will suffer starvation and pandemias in developing countries.”

Frank Pasquale, a law professor at a state university, wrote, “The key here is not that there is some predetermined path tech will take. Rather, current levels of inequality will be reinforced by robotization as more of these computers are used both to a) do present human-performed jobs better and b) suppress dissent or political action designed to better distribute the gains from technological advance. Think about Occupy Wall Street being dispersed on its first day by land-based robotic policemen and aerial LRADs. They will make the lives of the top 5% or so a virtual paradise, and will surveill and discipline the bottom 95% to keep them in line.”

A principal engineer for Cisco wrote, “Robotics will add a new twist to the global redistribution of manufacturing; if a robot can operate as cheaply in Detroit as in Shenzhen, why pay to ship materials and finished goods around the world? The social consequences will be driven by chronic underemployment and how we choose to manage it economically. Traditional unemployment schemes will not suffice. Some kind of negative income tax based system may be needed to ensure that everyone has enough to live on. Nevertheless a huge social and economic gulf will open up between those who work (even occasionally), and those who never work, and this will have dramatic political consequences.”

If we aren’t careful, increased income inequality and mass unemployment may lead to social unrest

Taken to their logical extreme, these trends—increased unemployment, widespread inequality, the emergence of a permanent poverty class—caused a number of experts to predict riots and other types of social instability in the relatively near future.

Vytautas Butrimas, the chief adviser to a major government’s ministry, responded, “Robotics and the assembly line have been with us for a long time. The jobs are still there but they now require more specialized training and skills. It seems there is a decline in general education while the elites continue to become more educated and more likely to get the high level jobs. The divide between the educated and less educated is growing wider. By 2025 we will have experienced our first major social unrest from this.”

The director of innovation a multi-country company aiming to tap into the gigabit Internet wrote, “I am participating in several international projects to develop agents and to bring about factories of the future (including hybrid factories with a mix of robots and blue-collar workers). I’m also participating in two international projects with major cities as partners, looking at ways of introducing enabling technologies such as the Internet of things. And I live in a city that has been chosen as a test site for replacement of some bus routes by AI-based vehicles. All of this leads me to set the job-displacement date earlier, 2020. And between 2020 and 2025 I expect a lot of social unrest, because insufficient attention is being paid to the needs of people displaced by technology. This will lead to a Winner Take All society, in which such workers can earn 10 or 20 times their current salary. Many of those citizens currently pay for part-time or full-time cleaners, gardeners, handy-people. Most of those local jobs will go (to judge from the people I know who already have robot cleaners and robot mowers). Very, very sad for the people affected.”

A technology risk and cybersecurity expert for a U.S.-based financial services association responded, “We have already observed how automation reduces employment, creates gaps in skills needed to be valued workers in multiple industries including the automotive industry. While it may be more efficient, leads to global trade, and moves complex supply chains, it also creates new challenges and problems for individuals and society. One of these challenges/problems is the gap in the skills and training that is necessary for workers to be valued. Another is increasing income inequality between those that have the valued skills and employed and those who do not and are unemployed or underemployed. Unless industry and government steps in to provide the necessary training, this could lead to greater political unrest.”

A browser engineer at Mozilla wrote, “Current trends indicate that the economy in its current form is ill-suited to support large numbers of low- and un-skilled workers. As more jobs become replaceable, I predict large societal upheavals as the gap between highly skilled (and highly paid) workers and a high proportion of partially, or totally, unemployed people continues to widen.”

A software engineer who works for a major U.S. technology company said, “I expect AI developments to take people by surprise by 2025. AI was oversold for so long and had so little success that people have forgotten about it. But I expect AI to be able to pass adult reading comprehension tests by 2020. There are large areas of the economy that will be affected by this and the skills needed to manage the AIs will be highly specialized and out of the reach of 95% of people. This will be very socially divisive. As power shifts from labor to capital, inequality will increase and social stability will decrease.”