Despite near-universal agreement that AI and robotics will make huge advances in the coming decade, the experts who participated in this survey are evenly split on the question of what impact those advances will have on human employment.
Some 52% expect these technologies to have a neutral to positive impact on jobs—that is, they disagreed with the statement that “networked, automated, artificial intelligence applications and robotic devices [will] have displaced more jobs than they have created by 2025.” When asked to explain their reasoning for feeling this way, they offered a number of rationales.
Throughout history, technological advances have produced as many new jobs as they displace—there is no reason to think that this long-standing trend will change now
The largest number of these experts used the historical link between technological advancement and employment levels (or lack thereof) to make their case. Celia Pearce, an associate professor of digital media at the Georgia Institute of Technology, was one of many to take this approach, saying,“Although they certainly shift and disrupt the labor ecosystem, if you look at the total net effect, history does not bear out the myth that technology replaces people. First, people make technology, and since technology becomes obsolete at an increasingly accelerated pace, the need for people who make it will only grow; second, people are required to maintain technology—technology is notoriously poor at taking care of itself; third, people are also required to assist other people in using technology; and fourth, most technology requires new labor forms.”
Rob Atkinson, president of the Information Technology and Innovation Foundation, responded, “I don’t think that automation is advancing any more rapidly than it has been. But in any case, automation has never led to fewer jobs in the economy in the past and never will in the future, for the simple reason that automation lowers prices which increases demand for goods and services, which in turn creates jobs.”
A Cisco Systems principal engineer wrote, “While the nature of work will change, there will be plenty of work. We don’t have many people making wagon wheels today, but we don’t attribute that to overall unemployment.”
A researcher for a major U.S. computer software and hardware company wrote, “This is a common misconception. If you read the economic literature, the evidence points to new technology creating new, different kinds of jobs. There is little to no correlation on unemployment and technological advancement.”
A long-time leader of technology development for the World Wide Web responded, “Since the Industrial Revolution an argument has been posited that technology will displace humans in the job market. It has not happened. Advances in robotics and AI are likely to provide new opportunities for human workers that are not realized at this time.”
Agustin Rossi, a PhD candidate at the European University Institute (EUI), Florence, Italy, wrote, “Disruption is not the same as displacement of aggregate jobs, as history since the Industrial revolution has shown us. It is safe to assume that technology will change how we produce goods and services, and that will probably mean that some people will lose their jobs in some sectors, but the economy will be able to create new jobs in different sectors that we cannot predict or imagine today.”
Brad Templeton, a leader with the Electronic Frontier Foundation and Singularity University, responded, “They will displace many jobs but the rate of job creation will outpace this. Why is uncertain—it always has, in spite of regular predictions to the contrary.”
John Wooten, founder and CEO of ConsultED, replied, “I don’t believe it will achieve a higher displacement over creation ratio by 2025. Current trends related to automation and business intelligence tools have surprisingly led to more job creation in the markets I have been involved with. For example, cloud computing has actually brought greater business necessity for hiring more IT persons, not less, as the implementation of ‘cloud’ affords IT personnel the ability to perform functions more critical to the organization as a whole.”
Niels Ole Finnemann, a professor and director of Netlab, DigHumLab Denmark, wrote, “The idea that use of digital media will reduce labor seems to contradict all former experiences. The labor market will change, but new jobs will be produced in the very same process, maybe in other places.”
Thomas Haigh, an information technology historian and associate professor of information studies at the University of Wisconsin, observed, “My position is basically historical. Over the past 200 years many technologies have boosted productivity by eliminating jobs; yet large-scale technological unemployment has never been a long-term reality. So it seems unlikely that new productivity technologies will destroy more jobs than they create. The economy will continue to shift towards personal services of a kind hard to automate.”
Carlos Castillo, a scientist working at a national research lab in the Middle East, responded, “In the long term the net effect of disruptive technologies such as the mentioned ones has historically been more prosperity, not job losses.”
An assistant professor at Radboud University Nijmegen in the Netherlands wrote, “Robots and computers have been around for quite some time. It has not displaced work. Workforces are just allocated in other sectors in the long run. In the short run it’ll displace jobs, but not in the long run. Robots, computers, AI will enable us to do more and more complex work, not the same work with less effort.”
Daren C. Brabham, assistant professor at the Annenberg School for Communication & Journalism, University of Southern California, wrote, “It is a long-standing sci-fi fantasy that someday our advances in automation/AI/robots will make human labor obsolete and allow us to live happier, healthier lives of leisure. That has never proven to be true—we work harder and longer in the U.S. now than we ever have, despite technological advances.”
Bill St. Arnaud, a self-employed green Internet consultant, wrote, “Robots and autonomous cars have been predicted and overhyped since the 50s. There will be niche applications but the decline in blue-collar manufacturing has been going on for decades because of productivity. Robots, AI, etc., are just an ongoing aspect of these productivity gains and do not represent anything special.”
A professor at a university in Wales, responded, “I find this hypothesis a scaremongering one. Technological growth of this kind is not like the sudden automation of a factory, it is an organic change in the nature of our work-lives on a global scale. An advanced Western nation like ours has an unemployment rate not significantly different from that it had in 1850 or 1950, barring the recent economic crisis. There is no reason to think that AI will shift employment rates significantly by 2025, or by 2055.”
Brian Butler, a professor at the University of Maryland, responded, “Most of the job losses are related to (a) shifting labor intensive work to parts of the world where labor is inexpensive (often enabled by communication and transportation technologies); and (b) relatively mundane automation (which is more reliable than AI). There is little evidence that large-scale shifts in employment are clearly attributable to AI/Robotics, despite fears of this occurring for the last 30-50 years. A key issue with this is risk management—how do we handle mistakes/errors/failures? The reality is that people are still better for this, if only because they can be held accountable/sued/fired/etc. That said, we probably will continue to see deskilling and reduced wages in many areas due to these technologies.”
Jobs will shift, as the same forces that displace certain jobs create entirely new types of employment—some of which we can only imagine today
Another large group of experts predicted that, while advances in technology might shift or displace jobs, these losses will be more than made up for by new jobs designing, building, servicing, and utilizing the same technologies that are displacing other types of work.
Gary Kreps, professor of communication and director of the Center for Health and Risk Communication at George Mason University, wrote, “The use of AI systems will supplement human systems and not replace them. For example, in health promotion I foresee the use of automated AI health education programs that will be used to provide additional channels for educating consumers about relevant health issues, but not replacing consultations with live health care providers. Additionally, there will be tremendous demand for experts to design, test, implement, and refine smart automated information systems, generating more jobs in the future.”
Bob Ubell, vice dean for online learning at New York University, wrote, “The history of technological advances can go either way. In some economic transitions, technological innovation can spur economic growth, creating vast new industries, with large new worker populations; but in other periods, technological advances can have the opposite effect, causing older industries to shed millions of workers. It’s far too soon to tell. In the meantime, for digital and other advanced technologies, the immediate effect of ‘creative destruction’ will more often be the loss of jobs. But, as history tells us, once initial destructive tendencies displace workers, some industries emerge with greater economic power beyond the overturned industry, creating large-scale new industries requiring an even larger labor force than the one displaced. But there is no guarantee that that will be the result in the industries identified.”
Nishant Shah, visiting professor at Leuphana University in Germany, wrote, “The question presumes that the advent of technology will not create new kinds of jobs. So while traditional jobs might be performed by machines and automated technologies, there will always be more jobs and more kinds of jobs that are catalyzed by the emergence of such technologies. The number of jobs is not merely going to decrease or deplete because the machines take over.”
Charlie Firestone, executive director of the Aspen Institute Communications and Society Program, responded, “Certainly there will be disruption with current jobs and more importantly, job functions. In the 2025 scenario, new jobs will be created by the robotic advances, and new functions will be created in the white-collar jobs of today. That is, there will still be a need for certain intermediaries but their functions will adapt to the needs of the new technologies and services.”
A general manager at Microsoft replied, “It is clear that advances in automation will eliminate some jobs, but it will create others as well as free up some resources that could be applied to other pursuits. I do not foresee a situation where we will have successfully automated humans out of work. On the contrary, I see a situation where we have greater need for higher-skilled workers who are comfortable with using and creating technologies.”
Joe Kochan, chief operating officer for U.S. Ignite, a company developing gigabit-ready digital experiences and applications, observed, “The design, programming, and creation of these devices and robots will still require, in 2025, more effort than they replace. These robots will, however, change the kind of work people do—robots will replace service and manufacturing jobs, but will open up more possibilities in tech and development.”
A pioneering academic computer scientist from Princeton University wrote, “Jobs will be displaced, but the improvements in efficiency and quality of life will lead to the emergence of more and better jobs elsewhere in the economy. The new jobs will be diffuse and it will be difficult to attribute the role of robots in facilitating their creation, but on the whole they will exist and will outnumber the displaced jobs.”
Luis Hestres, a doctoral candidate and graduate research assistant at American University’s School of Communication, wrote, “The emergence and spread of robotic advances and AI will inevitably eliminate the need for many jobs. However, it will also create new jobs that must be filled; for example, jobs that involve designing, manufacturing, maintenance and repair, and disposal after these devices are no longer functional. These devices will also probably create ancillary markets, much like the iPhone created the apps and accessories markets. The catch is that these jobs will require both more and better education than the U.S. currently seems prepared to offer its citizens on an equal basis.”
Jamie LaRue, a writer, speaker, consultant on library, technology, and public-sector issues, wrote, “The result of rapidly improving technology is not unemployment; it is a shift in employment. The lower level jobs will disappear. But new ones, designing and managing those systems, will grow exponentially.”
Rajnesh Singh, regional director in the Asia-Pacific region for the Internet Society, wrote, “Society has always evolved, and will continue to do so. Yes, some jobs will be displaced, but yet others will be created, and out of this will come further opportunities.”
A research fellow at the Global Cities Research Institute at RMIT University replied, “I tend to think technological ‘disruption’ is often overstated, and applied in too simplistic a fashion. There is considerable difference between the complete obsolescence of professions and associated ways of life under, for example, 18th century automation of the textile industry, and the various employment market adjustments that many contemporary innovations bring about. Self-driving cars and robots in particular are potentially transformative innovations in sectors that are chronically under-resourced today in developed societies, such as aged-care, rehabilitation and disability services. Hence the introduction of these technologies, even if pervasive, is unlikely to elicit massive change in any one sector, but rather is likely to continue an ongoing evolution towards high-skilled, specialized and ‘adaptive’ forms of labor.”
Estee Beck, a doctoral candidate at Bowling Green State University wrote, “In America, we have witnessed the shift from a largely manufacturer society to an increasingly software society, where capital and labor derives from the automations performed by computer algorithms. While we continue shifting from one source of revenue to another due to technological advancements, there will continue to be a need for people to oversee and work in the design, production, marketing, etc., of AI technologies.”
A law professor at Georgetown University and former U.S. Federal Trade Commission official wrote, “There is value to having machines making decisions and taking actions now taken by humans. AI is the new tool that may free humans from certain labor burdens—driving in traffic; lifting heavy objects; doing difficult computational functions—in exchange giving humans more time for even more productive labor. Every substantial technological breakthrough has been accompanied by a short-term drop in labor demand, then followed, as new applications developed, by periods of robust economic and job growth.”
Lillie Coney, a legislative director specializing in technology policy for a member of the U.S. House of Representatives, replied, “AI will be achievable because of advances in computing. However, it is still AI not human intelligence. There will be jobs to teach machines to perform their functions. People will work to perfect the technology to perform certain jobs, while determining the level of intelligence needed by each device. Creating something to perform mining that has the IQ of Einstein would be wasteful. Monitoring of technology to be sure that it is performing as intended will create opportunities.”
There are many jobs that robots simply will never be able to do, no matter how advanced they become
A sizeable number of respondents noted that there are many attributes—such as empathy, creativity, judgment, or critical thinking—that are uniquely human, and that technology may never be able to duplicate. As such, jobs requiring these skills will remain relatively immune from encroachment by automation.
Celia Pearce responded,“I actually see us moving away from AI and towards more crowdsourcing approaches. These tend to work better because it’s been proven when you throw a large number of human minds at a problem you can often get a better result than trying to get a computer to resolve it. Truth be told, computers are not very smart. All they are is giant calculators. They can do things that require logic, but logic is only one part of the human mind. Inspiration, creativity and intuition, meaning-making, storytelling and communication are all things that humans can do that computers will never be able to achieve fully.”
The dean/provost of a research university and former CEO of the California Virtual University wrote, “Intelligent devices calculate better than human beings, but they are not creative and don’t have judgment. They operate according to laws, and law has from the beginning needed human intervention to prevent it from behaving like an idiot.”
An anonymous respondent wrote, “Anything involving creativity, the arts, or nontrivial conceptual synthesis will remain the domain of human thought for the time being—I’ve seen no indication that machines could produce an award-winning screenplay or a work of true scholarly genius.”
Herb Lin, chief scientist for the Computer Science and Telecommunications Board at the National Research Council of the U.S. National Academies of Science, wrote, “Those areas in which human compassion is important will be less changed than those where compassion is less or not important.”
A university professor and researcher wrote, “I imagine that anything that can be better accomplished through a decision structure will be accomplished through AI entities, but I also see a growing value of the work that can better be accomplished by people. Detecting complaints is an AI problem. Sending the complaints to the correct customer service entity is an AI problem. But customer service itself is a human problem.”
The CEO of a professional not-for-profit society responded, “While I firmly believe we will have more AI and robots in our lives, none of those tech advances have judgment, and the human race and its enterprises will always value and need judgment.”
Fred Hapgood, a self-employed science and technology writer, responded, “Most of the people we interact with in real life are service people, and services are very hard to automate—especially the more expensive services. When you go to even a medium-quality restaurant in 2025, you will still be waited on by a human.”
Other respondents argued that our desire for human conversation and interaction will prevent many industries from going fully automated:
Deborah Lupton, a research professor on the faculty at the University of Canberra, Australia, wrote, “These technologies will displace some jobs, but they will also create others. Humans will always have the need for affective and embodied interactions with other humans, which can never be replaced by robots. This will particularly be the case in the context of healthcare, education, childcare and the care of the elderly. The attempted introduction of too many robotic devices may well lead to a backlash, in which humans who can provide care and education will become valued.”
Valerie Bock, technical services lead for Q2 Learning, responded, “Robotic servants will likely be part of the domestic and industrial appliance landscape. But I think we’ll attempt, and then give up on, machines as caregivers. We aren’t going to find a replacement for the human touch any time soon.”
An anonymous respondent replied, “There are always going to need to be areas that need a human’s touch. Even with the advancement of AI, I doubt they will get to a point in the next 11 years that will eliminate [the need for] human empathy and sensibilities.”
An administrator for technology-focused units in educational nonprofits responded, “People will have opportunities to have access and, as important, time to engage more types of information/activities/resources as they are freed from doing (or thinking about) ‘physical and mental basics’ for themselves. Relatively unchanged areas of life may cluster around aspects of living and working involving human needs for social, educational, economic, and political interactions that occur optimally (functionally) in the physical presence of other people—for example, working with babies or the elderly, or working with hands-on/minds-on aspects of peoples’ healthcare or educational needs.”
Political and sociological factors will prevent widespread elimination of jobs
A smaller but notable group of respondents anticipate that cultural and sociological factors—including regulatory inertia, liability fears, and public resistance to widespread displacement of jobs by robots and AI—will prevent new technologies from taking too big of an employment bite.
William Schrader, the co-founder and CEO of PSINet Inc., the first commercial ISP, observed, “Let’s start by imagining self-driving vehicles eventually working flawlessly and inexpensively in 2025. In that case, you imagine why anyone would need taxi drivers in 2025? Or truck drivers? Or limo drivers? Or pilots? Or boat/ship captains? Whether you call these white- or blue-collar jobs, if self-driving vehicles become reliable then how could we still have these jobs? In fact, why would we drive to the grocery store to pick up fresh food? We can already place the order on the Internet, the store picks what we want and (in 2025) the automatic self-driving truck would make the delivery to our doorstep (in some fashion). If self-driving vehicles are working flawlessly, and they spread across the planet within the next 25 years, then people will eventually forget how to drive. It is a concept like “rolling down the window”, after these rolling machines were replaced with electric switches for the past twenty years. People will still need to go places, and they will say ‘I am going to drive over,’ but perhaps no one will drive. But, the answer is no. By 2025, society will not allow free roving self-driving vehicles. It will take longer to have society embrace this change. And job loss will be one of the major contributing factors to the slow adoption.”
Liam Pomfret, a PhD student in online consumer privacy at the University of Queensland, Australia, thinks that government regulation will delay the widespread implementation of technologies like driverless cars: “I do not see robotic devices having much impact outside of the manufacturing sector (where they’re already firmly entrenched) by 2025. It’s unlikely that AI will see much penetration into markets by 2025, if only because government regulation regarding self-driving cars, etc., will be as slow to catch up with the changing landscape there as it has been with regards to Internet issues.”
A technology policy expert agrees, noting, “There will be a steady growth in this area, but it’s hard for me to say what its impact will be outside of certain niche sectors of the economy. Sure, certain factories will become highly roboticized, and I do think we will see self-driving cars in the next 5 years, but how long for market penetration to a point that it becomes more than a niche? How long for regulations to change that will allow some of these developments to operate within current conditions? That, I think, is more than 13 years away.”
An employee of the Network Information Center thinks that we are not prepared to live with the inevitable mistakes along the way: “Robotic advances and AI will obviously make tremendous headway by 2025, but there will be several highly visible screw-ups that result from these advances, which will generate widespread skepticism about how broadly AI can be used effectively. Note: these screw-ups will be the result of flawed decision-making logic that a reasonable human being would disagree with, which will create a significant backlash against an automated and programmed world.”
A technology developer and administrator wrote, “Forgiveness of human mistakes will continue to be higher than that of AI, as AI mistakes will be seen as systemic since the same mistake will be present in all versions of the software. Therefore AI and robotic use in more advanced applications will continue to be stalled in favor of simple robotics such as is used in manufacturing. The cost of human labor in many times will be lower than the larger scale robotics that could displace it.”
A program manager for an international nonprofit that promotes access to electronic resources in developing and transition countries replied, “No way, this is illusionary fear that comes with every technological innovation…it is naive to believe that businesses and organizations and people will have enough resources needed acquire most advanced tools. We can also expect some movement to stay ‘robot and AI free.’”
The technology won’t be advanced enough to widely displace jobs by 2025
A final group of experts simply feels that 10 years is not long enough for AI and robotics to become sufficiently advanced—or cost-effective—that they can replace huge portions of the workforce.
Chris Donley, director of advanced networks and applications for CableLabs, responded, “By 2025, no, I don’t think so. In this timeframe, I see robotics as primarily addressing convenience—allowing me to read a book while I commute to work, cleaning my house, or serving as a digital concierge. In this timeframe, robotics will primarily address things I would otherwise do myself, rather than pay other people to perform.”
Jane Vincent, a fellow at the Digital World Research Centre, responded, “These are aspirational aims that by 2025 will have had a major impact on some areas of work and society but will by no means be the norm. Perhaps the biggest change will be in automated management of public services reducing human interaction at the first line of enquiry.”
Uta Russmann, a professor of strategic communications based in Europe, responded, “Even though technological developments are developing faster each decade, it will take more than another 10 years until such a scenario will be reality. In 10 years the use of robots, digital agents, and AI tools will still be too expensive for everyday use and broad economic use.”
John Lazzaro, a research specialist and visiting lecturer in computer science at the University of California-Berkeley, wrote, “As an engineering community, we’ve been working on robotics and AI for a long time. The rationale behind today’s optimism is that with every process technology generation, Moore’s Law brings us closer to having enough computational resources to solve these problems well. The adult human brain has about 500 trillion synapses, but enough racks full of GPUs fabbed in the fully-scaled 5 nm CMOS process may be able to simulate the pattern recognition abilities of those synapses, at the right level of abstraction for solving engineering problems. This is the basic line of reasoning. I’m skeptical, because don’t think we will have identified the full complement of ideas yet to write the algorithms that would run on the GPU cluster…Someday I believe we’ll have those ideas. But 2025 feels too soon—it takes decades for fundamental ideas like quantum mechanics to be fully worked out by a research community, and so we would know it was coming by now if it were ready to deploy by 2025.”
Stephan Adelson, president of a technology consulting company, predicted, “Robotic advances will displace jobs, but at a slow rate. 2025 is not very far off and the technology needed to create serious disruption in employment will take more time. Innovations such as self-driving cars will take a great deal more time than small innovations such as we see now.”
Peter S. Vogel, Internet law expert at Gardere Wynne Sewell, LLP, replied, “I have been following artificial intelligence (AI) technology since 1971, and in spite of today’s great innovations I am not optimistic that AI will displace all that many applications or robotic devices by 2025.”