The concept of a job (that is, work done in exchange for payment) is relatively new in human history. We spent a lot longer foraging and hunting for food. Before we invented working for pay, we invented free labor in the form of slavery. In a way, working for payment was an improvement over being a slave: freedom from being someone’s property. But what started as freedom ended up becoming another kind of slavery in which those who accumulated more resources than the rest of us claimed the right over our time, our muscle power, our brain power, and ultimately our choices. We kept inventing mechanical devices to reduce the drudgery so we had the “convenience” of time to do other things, like having wars and writing poetry, but also building even better and faster machines. Some of us became “middle class” and created ideas like democracy, higher education, the symphony, world peace, and summer vacations. As children of the middle classes we started spending longer and longer time in school which allowed us to invent more machines, and “languages” through which we could talk to the machines, and languages with which they could talk to each other. We invented the Internet, Facebook, Instagram… As we played endless numbers of computer games shooting virtual birds and feeding virtual cows, we did not realize that our time was losing value because what was done by the purchase of our time now could be done for free by the machines that we kept inventing. Scores of us started losing our jobs and our middle class standing. And it was not because our jobs were going to places of cheaper labor because they were also losing jobs. All of a sudden, because a few centuries is not very long in human history, we realized that we were in a world of 9 billion people with only a fraction of it in the protective dome of a good life and the rest of us barely eking out a living outside.
(Click here for Part II)
Wake up call: one study many reactions
In September 2013, two Oxford professors, Carl B. Frey and Michael A. Osborne, published an article that analyzed susceptibility of jobs facing replacement by automation. Impact of technological change on jobs is not new but the result the study presented was a wake up call. The analysis of 700 job categories in the US showed that 47% of them, or almost one out of every two jobs, could be replaced by automated systems by the mid-2030s. Jobs in the service, sales, transportation, extraction, farming and administrative sectors were the most vulnerable. Jobs that involved creativity, social interaction, management and financial know-how, and software engineering faced a lower risk of automation. The possibility of half the working people (in the US) becoming unemployed within less than two decades freaked everyone out although this possibility had been written about earlier by other economists.
There were many reactions to the Oxford study. Within weeks, Slate magazine wrote about weaknesses of the study. A few months later, the Economist argued that technological progress “squeezes some incomes in the short term before making everyone richer in the long term,” but “has never previously failed to generate new employment opportunities.” This view was embraced by most businesses, management consulting firms, those creating the automated systems and the investors.
In January 2015, Deloitte published a study that reviewed historical job trends in the UK from 1871 to 2014. It identified jobs categories that have been shrinking (metal workers, secretaries, farm workers) and those growing (management consultants, financial managers, hairdressers, nurses, teachers, actors, bar tenders). Deloitte emphatically argued along the same lines with the Economist: technology has always been a net creator of jobs.
A McKinsey study in November 2015 went beyond the cost-cutting value of automation and looked at its implications for the content, context and future of jobs. This study found that automation was not eliminating but redefining jobs, and that fewer than 5% of occupations could be fully automated. It also found that 60% of occupations could have a third or more of their constituent activities automated, leading to significant job redefinition and a transformation of business processes. McKinsey was optimistic in how it depicted the impact arguing that with the routine parts of a job automated, people would have more time to focus on higher value aspects of their jobs.
In January 2016, the World Economic Forum (WEF) released an extensive study on automation and job loss for its annual meeting of business bigwigs in Davos, Switzerland. The WEF study was the first to take a more global perspective: it focused on major industrial countries besides the US and the UK, plus five emerging economies (China, India, Mexico, South Africa and Turkey), and aggregate data on four countries each from the regional groups of Association of South East Asian Nations (ASEAN) and the Gulf Cooperation Council (GCC). The analysis was based on survey responses received from 371 companies with a combined 13 million employees in 9 industrial sectors. WEF concluded that, by 2020, seven million jobs would be lost to automation while only two million new jobs would be created. The study made it clear that this time around, technology was not a net job creator and unemployment due to automation and cognitive technologies was markedly different from previous industrial cycles
In the flurry of reactions there was hardly any attention placed on the developing countries until studies by the World Bank and the International Labour Organization (ILO) were published in 2016. The Bank focused its 2016 World Development Report (WDR) on the digital dividend and the unemployment impact of automation across the developing world. The report estimated that impact of automation would be a 60%, or higher, loss of jobs in most developing countries. Estimates of impact in specific countries included 69% for India, 72% for Thailand, and 85% for Ethiopia.
The ILO study focused on Asian countries (ASEAN members) at specific regionally important industrial sectors (automotive, electrical/electronic, textiles/clothing, business process outsourcing, and retail). The ILO’s conclusions echoed that of the Bank. For example, in the textiles/clothing/footwear sector, which currently employs 9 million workers in the region, 85% in Vietnam and 88% in Cambodia would be at risk of being replaced by sew-bots and other automated systems.
These figures made the 47% unemployment, that Frey and Osborne had predicted for the US, seem manageable. The developed countries generally have higher literacy rates and better education infrastructure to retrain and re-align the workforce with the emerging jobs market. But how does a country like Ethiopia with an average literacy rate of 39% handle a quadrupling of unemployment? Job loss of 60% and higher will be devastating in countries where there is already high unemployment, and widespread poverty.
Automation and jobs: loss or gain?
The fact of today is that anything that can be automated will be automated: because it saves money and time. Faster and cheaper are desirable for obvious business reasons.
The most automatable jobs, or job tasks, are those that are routine, repetitive, often manual, and requiring few or no cognitive skills such as creativity. Why? Because of the nature of computer programming underlying automation. Anyone who has taken even a basic coding class would be familiar with the loops of instructions that can run forever, or until you insert a line instructing a stop. So if you have determined the repetitive and routine tasks of a job, all it takes is to write loops for them and run them as a chain. Voilà! You have automated the job, or at least parts of it.
Throughout industrialization, we have been determining the routine-repetitive jobs or job tasks, for efficiency and process management purposes. This not only gave us the assembly lines but also primed many jobs for automation.
Educated guesses of which jobs, or job tasks, automation will eliminate usually starts with a new technology, say the driverless car, and extrapolate its impact. While the immediate impact of the driverless car is on the drivers of all kinds of vehicles from taxis to trucks, it does not end there. Jobs that flourished around the car are also affected such as traffic cops, meter maids, traffic court judges, parking lot attendants, driving schools, and valets.
The jobs-lost lists that many futurists, business analysts, technologists and others put forward include similar categories: drivers, retail sales people, waiters, and workers in the manufacturing, agriculture, construction and mining sectors. Their lists gradually expanded to those that were previously considered safe, such as basic managers and office administrators, accountants, and legal clerks. There were also predictions of jobs gained in business and financial operations, (upper) management, computer and math professionals, architecture and engineering, and education-training. The general assumption was that jobs that use cognitive, creative and social skills would be safe and in high demand in the future. But this assumption already needs a re-think.
For example, writing is considered creative and non-routine. Yet there is already software that can write prose and poetry well enough that human readers cannot distinguish whether the author is a human or software. Acting is another creative occupation yet Synthespians, digital clones of real actors, is already a fact though not yet used at scale. What about music? Google Magenta is software that composes original music and Jukebox, an online software, generates new music based on choices of genre, musical period, and mood.
iCEO, developed by the Institute for the Future, is a digital management prototype. Harvard Business Review tested the prototype to manage a research project for a client: the project took a fraction of the time it would normally take with human leadership and produced a high quality result. A 2015 study by Gartner predicted that by 2018, more than 3 million workers globally will be supervised by a robot boss. We do not yet have a robotic Bill Gates or Jack Ma but many aspects of management are already being automated fulfilling this prediction. Attitude surveys of employees, reported by Forbes, show that two thirds of workers would trust a robot more than a human boss because they feel the robot boss would not be biased and erratic. On the other hand, you cannot get compassion and empathy from a robot though you may get it from your erratic human boss.
A few books of note
Many smart people are monitoring these technologies, writing best sellers on what to expect, and on how to cope. Some are optimistic and some pessimistic about the future. Generally, those who are optimistic seem to be those who own a piece of these technologies. Here is a quick list of some of their books, in no particular order :
- The Singularity is Near (2006) by Ray Kurzweil the inventor, investor, futurist and all around genius. He is very optimistic about the future, impatient for the arrival of the singularity where machine will merge with human creating a new species. In case he does not make it to the arrival of the singularity, he has arranged to be frozen and revived after singularity is reached. I read this book long after it was published and regret that I did not note it earlier.
- Raw Deal (2015) by Steven Hill who is realistically pessimistic with a focus on workers and the future of society, exposing the exploitative aspects of the emerging sharing economy that is built on automation and artificial intelligence software such as the likes of Uber, Task Rabbit and AirB&B. Hill demonstrates how these new companies make huge profits but redefine work for the worse as they convert people into temporary contract workers without any benefits or much respect for their labor.
- Only One Thing Can Save Us (2014) by the labor lawyer Thomas Geoghegan is a sobering book that focuses on the vast social implications of automation and other emerging technologies. He calls for a revitalized labor movement in which unions use these emerging technologies for their benefit to mobilize, educate and organize.
- Abundance and Bold (2012 and 2015 respectively) by Peter Diamandis (and Steven Kotler). Both books are giddily optimistic accounts of a future full of opportunity, with people freed from boring work as well as physical limitations including aging and death. The authors serve a bit of an attitude towards those who are not as excited about this future. Diamandis is an exponential technology guru, aerospace engineer and medical doctor, not to mention founder of the X-Prize Foundation and many other enterprises. Diamandis and Kurzweil, share the optimistic view of the future and collaborate, including as co-founders of the Singularity University. (SU provides training programs as well as summits and other meetings which are not affordable by those the future is leaving behind: the executive training program price tag is $15,000).
- The Second Machine Age (2014), by two MIT professors Erik Brynjolfsson and Andrew McAfee, is neither pessimistic nor optimistic befitting their academic profile. Still some have labelled them “doomsayers” for observing how more and more human labor is being replaced by machines while humans have no where left to go. They also show how the new technologies are creating a winner-take-all economy that will exacerbate already off the chart inequalities.
- Rise of the Robots (2015) by Martin Ford, the software entrepreneur leaning towards the optimistic side with his belief that automation will lead to an overhaul of the current economic system and people will have more productive lives with their basic needs covered through a guaranteed income.
- The Glass Cage: Automation and us (2014) by Nicolas Carr, one of my favorite authors in this field. He has written several best-selling books as well as many articles for the NY Times, Atlantic and elsewhere. Carr builds his utterly and personally relatable arguments meticulously, logically and yet engagingly. He is neither a technology pessimist nor optimist but stands in the corner of human-ness. His Rough Type is a good introduction to his work.
What to expect and how to cope: couple of scenarios
Many smart people have proposed future scenarios of the age of automation or suggested ways to cope with the changes ahead. Here are two examples:
Example I
Peter Diamandis, the exponential technology guru mentioned earlier, sees four possible scenarios ahead:
- technology leads to net positive increase in jobs and all is well;
- jobs are lost but society adopts (jobless become artisans, they barter, they create cooperatives);
- near term jobs loss, significant gains in the long term all is well eventually; and,
- jobs are lost and not recovered leading to social rebellion.
The first scenario is what Diamandis and other automation optimists want the most. But, how likely is it? While new technology has historically created more jobs than it destroyed, studies on current technology and jobs, such as the one by the World Economic Forum, point to a net loss of jobs and argue that the impact of today’s new technologies is different than the impact of technological changes of the past.
Growing inequality worldwide supports the fourth and the least desirable scenario. Inequality is already leaving behind millions of people. More will likely join the ranks as they are made redundant by an automated economy and lack access to retraining for the new jobs that may emerge. The late Stephen Hawking and Dr. Brynjolfsson of MIT, among others, have acknowledged the growing inequality because benefits of the new technologies are not distributed fairly.
The third scenario depends on how “long-term” or “eventually” are defined. How long is long-term? Is it a generation, two generations, longer? The longer the time frame, the less likely “all will be well eventually”. Those who are receiving the benefits of new technologies can afford to wait. But what about those who cannot afford the wait time?
The second scenario rests on the human skill for adaptation, which has served us well over the millennia. And there are stories of communities creating cooperatives, bartering services and goods, and becoming artisan cheese makers. But adaptation requires time just as you need time for your eyes to adapt from light to dark. If the speed of change is not faster than the time available to adapt, then we can adapt successfully. But with exponentially advancing technologies, our adaptation skills fall short.
The inequality context is largely left out of these four scenarios which is disturbing. Inequality and its socio-political and economic impacts need to be front and center of the automation and jobs debate not dismissed. Otherwise we are looking at a future in which a small group of elites are living longer, healthier and more luxuriously in their robot serviced and protected gated compounds while the rest are left to eke out a living by bartering and making artisanal cheese at best. This to me sounds like a return to the feudal system except with robots.
Example II
Marc Andreessen, a technological progress optimist, creator of Netscape and a serial investor, proposes, on his blog, three interrelated approaches for managing the future:
- Focus on increasing access to education and skill development, which itself will increasingly be delivered via technology.
- Let markets work (this means voluntary contracts and free trade) so that labor can rapidly reallocate into new fields and jobs.
- Create and sustain a vigorous social safety net so that people are not stranded and unable to provide for their families. The loop closes as rapid technological productivity improvement and resulting economic growth make it easy to pay for the safety net.
More education is an approach with near universal appeal. Unfortunately, it has inherent problems. First education and retraining take time to bear fruit and not helpful to people who need retraining now. Second, most education systems around the world do not teach learners the kind of non-routine cognitive skills such as creativity, curiosity, and collaboration which future jobs will demand. Neither do they offer programming-coding skills which are becoming as basic as reading-writing-math have been. Significantly overhauling existing education systems will need significant resources. And lastly, there is the matter of teaching programming or machine learning skills to people who are functionally or literally illiterate.
The second proposal of letting markets work sounds hallow in a world in which fewer and fewer people believe in the beneficence of the markets left alone. Especially the young people around the world, see that it is the unregulated markets that brought them to dwindling jobs, growing inequality and a murky future. Besides, market principles emphasize cost reduction and does not care if that is achieved by replacing people with machines, just as it did not care that slave labor was the engine of sugar plantations. Rather than letting markets “do their thing”, wouldn’t it make more sense to regulate and structure them better so they do not trigger undesirable social outcomes?
Andreessen’s third proposal is interesting and laudable in its concern for the future of society and an underlying desire to protect it by a ‘vigorous social safety net”. Although the term is not used, the idea evokes the concept of universal basic (or guaranteed) income which is increasingly more people in the tech world are coming to support. I personally like the idea a great deal, but also known that it begs many questions. (More on this later)
Implications for Society
The geeky, sci-fi loving, future-oriented me is optimistic about the promise of these developments. But the political scientist, political economist, and international development expert side of me is quite pessimistic as I cannot get over the political, economic and social impacts of what is coming and its impact on innocent people. I know it will be the poor and disadvantaged who will suffer most.
A key reason for my pessimism is that many researchers agree that the unemployment result of digital technology is very different from that of mechanical technology. This is well argued and presented by Douglas Rushkoff in his 2016 book Throwing Rocks at the Google Bus: How growth became the enemy of prosperity.
Technological unemployment, a term attributed to John Maynard Keynes, has been part of our history. We know how horse carriages were replaced by cars and sock knitters were replaced by knitting machines. We know that the stablemen, the carriage drivers and sock knitters lost their jobs but these technologies created new demands and needs: for roads, tire makers, gas stations, colorful sock designers …Hence there were indeed more new jobs than jobs destroyed. This is the story that people optimistic about our automated future repeatedly tell. But often they omit a critical detail: in the past productivity gains from the new technologies were reflected on the wages as well as in the profits. This is not the case with digital technology – productivity gains from these technologies do not lead to higher wages. In fact, a great decoupling of jobs, wages and productivity has been in effect since the late 20th century.
There are important political implications of these changes in that we may be forced to part with the values that we have came to take for granted such as freedom, rule of law, equality and democracy.
Scholars have long pointed to the direct relationship between democracy and the middle class (examples here and here) . The origin of this link is Aristotle’s Politics in which he writes “the best political community is formed by citizens of the middle class, and that those states are likely to be well-administered in which the middle class is large, and stronger if possible than both the other classes.…” (Book 3, part XI)
A strong middle class has the pull to support rule of law, protection of property, and personal freedoms as these are beneficial to their well-being and prosperity. Decline of the middle class is historically associated with reactionary politics and fascism as shown by Martin Seymour Lipset in his 1981 work “Political man: the social bases of politics”.
If the on-going net loss of jobs is pulverizing the middle class in most industrialized countries, can we depend on the survival of rule of law, legal equality of citizens, individual freedoms, participation of citizens in decision making and accountability of the government? The rich upper class supports these middle-class based values because they need the middle class to manage their businesses and need the poor lower class to provide the necessary labor. If the middle-management and workers are made redundant by machines, what incentive does the rich class have to support, financially and politically, the democratic institutions? And if the democratic values erode, how will our political systems change? What type of political systems might emerge? Do we end up in a “digital feudalism” in which very few have all the money and the power, and the rights and livelihood of the rest are dependent on the whims of the few?
There are hints of a different middle class forming in some emerging economies. For example, a 2017 study, shows that the middle class in China, while expressing support for democracy and social justice, is “prepared to be subservient to an authoritarian state in return for economic security and socio-political stability.” Since he middle class is growing in emerging economies while declining in the industrialized world, future state of democracy and its institutions may be different from those that emerged in Europe during the industrial revolution.
Leave a Reply