For 200,000 years, humans have invented tools to make work easier, and at times, enable work that we could never do before.  Twelve thousand years ago, a technology called Agriculture gave us abundant food supplies to fuel the growth of civilization. Just 250 years ago, we harnessed steam and steel to multiply human strength and unleashed the Industrial Revolution.  Less than 80 years ago, we invented digital computing machines and started a new phase that continues to drive change at an ever-increasing pace.

There is no doubt that technology has improved our lives and work immeasurably since the stone age. As with the Luddites in the 19^th century, much is being said today about the dark side of technology and its impact on workers.  As we enter the current phase of the industrial revolution – Industry 4.0 (Schwab, 2015) – a fusion of physical, digital, and biological technologies, we read each day how these new technologies are degrading jobs, leading to lower compensation, and ultimately making workers obsolete and unemployable. Some predict future technology will leave large groups of people unemployable, excluded from the economy ^[1]. It appears the Worker is under siege by this wave of new technology. But is this true?

Prometheus Endeavor has reviewed current research on the impact of technology on the workforce with the goal of offering practical advice to improve outcomes for workers, employers, and society. This document summarizes our key findings and provides a list of questions for further attention.

Historical perspective

The Industrial Revolution began in the late 1700s with the invention of the steam engine. Since then, it has progressed through new phases, spurred on by new technologies. Currently, we are in the Fourth phase, Industry 4. See Table 1 below for a summary of the four phases.

Table 1. Phases of the Industrial Revolution

Each phase of the Industrial Revolution was initiated by a major invention leading to multiple transformational technologies with sweeping consequences throughout society.

For example, the internal combustion engine led to the birth of the automobile and aviation, which, in turn, had substantial impacts throughout the economy, the workforce, and our way of living. Likewise, the transistor, and its subsequent miniaturization into integrated circuits, enabled all the digital technologies we enjoy today.

The first two phases of the Industrial Revolution automated manual work. With the invention of the digital computer, automation extended to cognitive tasks in phases 3 and 4, touching every aspect of our lives.

Some argue that we are still in phase 3 and that phase 4 is an extension of earlier inventions instead of a major new one. Nevertheless, Industry 4.0 features ubiquitous transformational technologies such as Artificial Intelligence, Robots, Big Data, and the Internet of Things. This phase can be divided further into three waves (PWC UK, 2018):

• Wave 1: Algorithmic wave (to early 2020s) Automation of simple computational tasks and structured data analysis, affecting data-driven sectors such as financial services
• Wave 2: Augmentation wave (to late 2020s) Dynamic interaction with technology for clerical support and decision making. Also includes robotic tasks in semi-controlled environments such as moving objects in warehouses
• Wave 3: Autonomous wave (to mid 2030s) Automation of physical work and manual dexterity, and problem-solving in dynamic situations requiring responsive actions, such as transport and construction.

All phases have also created and displaced jobs across a broad spectrum of skills and compensation levels. Phase 4 automation is targeting mainly unskilled blue- and white-collar workers. However, every worker, including higher-paid professionals such as doctors, lawyers, and hedge-fund analysts, will also be affected.

Over the years, despite its disruptive impact, technology has led to significant net increases in the workforce’s overall employment, as demonstrated by the pre-pandemic U.S. unemployment rate of 3.5% – not seen since the late 1960s. There is no guarantee, however, that this trend will continue. In the past forty years, the number of jobs added to the U.S. economy has consistently declined from an annual rate of 1.7% to 0.6% (U.S. Bureau of Labor Statistics, 2016). 

It is not clear whether this rate will turn negative in the future, even if the economy continues to prosper.  Predicting how many jobs technology will displace in the near term is much easier than estimating which jobs and how many it will create.

Reasons companies automate

Automation helps companies remain competitive, and saving money is the primary reason companies automate. These savings take different forms, such as increased productivity (i.e., lower unit cost), reduced labor costs, and physical space savings (GAO, 2016).   The U.S. tax system also incentivizes companies to automate, as the tax rate for hardware and software expenditures is approximately 5% versus 25% for labor (Acemoglu and Manera, 2020).

Non-economic reasons for automation include increasing job quality and worker safety by reducing tedious and dangerous work and maximizing workers’ value-added by complementing and augmenting their capabilities.

Also, sectors like agriculture and online sales fulfillment centers turn to automation to avoid worker shortages during seasonal peaks.

However, these are not the only reasons or even the most important reasons for companies to automate.  Companies across industries are automating to accelerate and improve research and development and the quality of their offerings and customer interaction.

The current Pandemic has also introduced a sense of urgency for organizations and people to operate remotely and find new ways of bringing their products and services to their customers (McKinsey & Company, 2020).

Work automation potential

According to McKinsey, by 2030, 95% of all U.S. jobs will be impacted by technology (Manyika et al., 2017).  This impact represents 160 million workers who will be touched by technology to a greater and lesser degree.  How will jobs be affected?  First, we need to keep in mind that technology does not automate jobs but rather the multiple tasks that make up a job. When most tasks comprising a job are automated, the job is ripe for elimination, with any remaining manual tasks subsumed into other positions.

Repetitive physical and cognitive tasks are the most susceptible to automation, such as crop harvesting, manufacturing assembly work, and insurance underwriting.  But new technologies like A.I., neural networks are now taking on tasks involving the analysis and identification of patterns in large volumes of data.  For example, analysis of crime patterns for deterrence or identifying malignant growths in x-rays and skin images.

Unpredictable tasks and tasks requiring dexterity and cognitive skills are less susceptible to automation. These include jet engine repair, carpentry, the arts, counseling, and managing people.  Categories of skills which are currently most difficult to automate with current technologies are listed below in Table 2:

Table 2. Skills that are less susceptible to automation

Automation impact on the workforce

As we saw earlier in Table 1, each phase of the Industrial Revolution has affected different segments of the workforce. In Industry 4.0, virtually every job has tasks currently, or in the foreseeable future, benefitting from some form of physical or cognitive automation.  McKinsey estimates that while just less than 5% of occupations are fully automatable, over 90% of occupations are partly automatable.  This potential may impact 400 million full-time equivalent workers worldwide, with 23 million in the United States by 2030 (Manyika et al., 2017).

Oxford Economics concluded that, by the same year 2030, around 8.5% of manufacturing jobs worldwide could be displaced by robots^[3] (How robots change the world, 2018). Their econometric models estimate that a robot replaces 1.6 jobs on average, and, surprisingly, 2.2 jobs in lower-income areas and 1.3 in higher-income areas. Robots also impact the service economy, including healthcare, retail, hospitality transport, construction, and farming. Of course, humans in service occupations that require empathy, creativity, and social skills will be more difficult to replace.

Let us turn our attention to the types of workers impacted by automation, currently and in the foreseeable future.  As we discussed earlier, each phase of the industrial revolution has impacted different segments of workers. Now in Technology 4.0, occupations with the highest probability of automation are at the bottom of the compensation and educational ranges (Osborne, 2013). 

These vulnerable jobs include workers in office and administrative support, service, sales, manufacturing, transportation, and material moving. A recent GAO report to Congress confirmation this demographic perspective of workers in jobs susceptible to automation, as seen in Table 3 (GAO, 2019).

Table 3. Demographics of U.S. workers in jobs susceptible to automation, 2016

Source: GAO, 2019

Looking at the table by column, Of the 137 million workers in 2016, 43% were in jobs susceptible to automation. In contrast, 61% of jobs held by workers with only a high school degree or lower were at risk. Viewing it by row, while only 11% percent of workers with graduate degrees were susceptible, 89% were safe from automation. Both observations emphatically show the relationship between education and job exposure to automation. Likewise, Hispanic worker jobs are substantially more susceptible to automation than other ethnic groups.

We have discussed the types of work and workers that are more likely to be impacted by technology in the foreseeable future. Table 4.0 presents a sample of the projected growth and decline of specific job categories between 2016 and 2030 in the U.S. This table was summarized from McKinsey’s 2017 report (Manyika et al., 2017), which includes a complete list of job categories and their projections for several world regions.

These projections attempt to reflect the impact of two forces that result in job growth or decline. Technology and its impact on job displacement is one key force. Demand for specific types of products and services requiring these jobs is the other force. The report models six potential sources of new labor demand, which may lead to job creation by 2030:

• Rising incomes and consumption, especially in emerging economies, increasing overall consumption.
• Aging populations demanding increasing levels of health care and personal services
• Development and deployment of technology, leading to new technical job
• Investments in infrastructure and buildings, calling for architects, engineers, carpenters, and other skilled tradespeople, as well as construction workers
• Investments in renewable energy, energy efficiency, and climate adaptation, leading to the creation of tens of millions of new jobs
• “Marketization” of previously unpaid domestic work, such as cooking, childcare, and cleaning, currently performed by mostly women who are joining the labor market in growing numbers.

Table 4. Selected employment growth and decline by occupation projections for 2016 – 2030

SOURCE: U.S. Bureau of Labor Statistics; McKinsey Global Institute analysis

Workforce automation dynamics

Transformational technologies have a broad impact across industries and work categories, automating many tasks across many jobs, creating substantial employment opportunities while eliminating relatively fewer jobs. The auto industry, for example, created 7.5 million jobs between 1910 and 1950 and eliminated only 623 thousand jobs (Manyika et al., 2017). In fact, these technologies affect our entire society and the way we live (e.g., the steam engine, electricity, the internet).

At the other end of the technology impact continuum, incremental technologies have a narrower impact by automating only specific tasks within fewer industries and practices, such as autonomous excavators and automated lab analysis tools. However, several incremental technologies working together can have a transformational impact on interrelated business functions.

Consider an order fulfillment center from Amazon where hundreds of robots scan, pick, label, sort items, and box them into parcels to ship to customers. All these tasks are orchestrated by an advanced A.I. system^[4].  Each of the machines involved in this complex operation can be considered incremental, but all of them in concert are transforming the workings of warehouses, distribution centers, and the entire supply chain.

Technology can affect jobs and workers in three ways, depending on its transformational or incremental nature and the tasks it automates:

Job creation. Technology creates new jobs in new job categories as well as existing ones. As an illustration, the personal computer – a transformational technology –  between 1970 and 2015 created 15.8 million jobs in the U.S. while eliminating just 3.5 million (Manyika et al., 2017).

Jobs created by technology fall into three broad categories:

• Direct jobs are those required to design, build, operate, and maintain the new technology.  For the personal computer, these include jobs in semiconductor design, manufacturing, and software engineering.  Additionally, larger numbers of workers in support industries like Microsoft and Oracle provide users with software and support services.  Overall, personal computers added 3.5 million direct jobs.
• Indirect jobs are those that apply new technology to create value within enterprises not directly involved in the provision of the technology.  For the personal computer, these positions represent another 12.2 million net jobs in occupations such as computer scientists in finance, manufacturing, business services, and a multitude of industries using P.C.s
• Macroeconomic multiplier effect jobs. Looking at the overall economy, we need to consider the macroeconomic multiplier effect (MME), by which gains in one form of economic value causes economic stimulus in related areas and beyond (e.g., professional services, finance and insurance, and leisure and travel) – translating into new jobs (Wright, 2017).  To illustrate this effect, 3D printing (3DP) has evolved in speed, size of objects they can fabricate, and types of materials they handle at lower costs.  This technology is becoming an important part of various manufacturing operations ranging from turbine parts to sneakers. These advances are allowing manufacturers to bring home some critical components currently contracted from offshore providers.

  AT Kearney predicted that, in ten years, this might result in the addition of 2 – 3 million direct and indirect jobs in the U.S.^[5]. They also estimate 1 – 2 million jobs will be added from the MME of this technology by applying an MME factor of 1.8 (Monahan et al., circa 2017).PwC UK analyzed the macroeconomic impact of artificial intelligence and estimated that by 2030 GDP will be 14.5% higher than in 2016 (PwC, 2018). Translating this growth into our three categories, roughly 45% or 1.8 million new jobs will be direct and indirect, and 55% or 2.2 million will correspond to the MME^[6]. For this technology, the MME factor is a more conservative 1.2.

Job transformation. Jobs that are partially automated remain, becoming more productive and effective than before. McKinsey (Manyika et al., 2017) estimates 95% of all jobs affected by technology fall into this category.  For the worker, this transformation can either improve or diminish the content of their job.

Jobs are improved when technology augments the worker’s physical or mental capabilities.  In other words, when the machine supports workers in performing their tasks better and with less effort.  Examples include robots assisting in a surgical procedure, 3D printers supporting designers in prototyping complex parts, A.I. aiding advertisers in media selection.

Jobs become diminished when automation takes over core tasks, relegating workers to a support role where the worker must support the machine doing its work.  Here, examples include warehouse automation, where workers handle tasks not yet automated, such as boxing and handling odd-shaped objects. In this case, the machine sets the worker’s pace and controls his performance, even going as far as firing poor performers (Lecher, 2019). 

Job elimination.  When most tasks comprising a worker’s job are automated, the work goes away. As jobs are transformed, becoming more productive, fewer workers are required to carry the same workload, and the remainder of workers lose their jobs. Findings from research of 28 industries in 18 OECD countries indicate that in the last 50 years, automation has increasingly focused on job replacement instead of job enhancement (Autor and Salomons, 2018). This trend leads to a growing number of displaced workers, facing four options:

1. Move to a higher-level job (e.g., supervisory, design, repair, installer). This path is a rare but feasible possibility for high-performing workers who get the opportunity, are adaptable, and show initiative
2. Find similar employment elsewhere in another company that has not yet automated their job category. This option will probably be at some loss of compensation.
3. Find new employment in a different occupation, at lower pay, most certainly below their competency level.  These workers make up the growing number of underemployed and frequently need to work two jobs to match their previous living standards. In the past 30 years, many assembly line workers and middle managers have lost their jobs to automation and had to find employment in construction, food preparation and delivery, and retail
4. Becoming discouraged and join the ranks of the unemployed. Depending on age, some may opt for early retirement.

Most workers finding themselves unemployed are forced into lower-paying jobs. The result is the stagnation of average hourly worker compensation. As overall productivity continues to grow, these underemployed workers share an ever-decreasing piece of the economy’s pie, as depicted in Figure 1 (Lawrence, 2015).

Figure 1. Real hourly wages and output per hour

The blue curve shows compensation for production and non-supervisory workers in the U.S. The red curve reflects all workers, including skilled and supervisory workers, whose wages have grown more rapidly than production workers but still lag the overall economy’s growth.

In addition to technology automation, other forces contribute to this growing gap, such as globalization, immigration policy, public policy, and political inaction.  History may hold a lesson based on a similar wage gap (Wikipedia) in the early 1800s that led to powerful social movements, labor unions, and Marxism.

The pace of automation

Technology automation continues its relentless advance on many fronts of human activity. Almost daily, we learn predictions of the imminent wholesale elimination of workers in many areas with high labor content: autonomous vehicles replacing drivers; restaurant automation eliminating managers, order takers, and cooks; robots replacing factory workers and warehouse operators (already discussed), and A.I. displacing financial services employees.

While these predictions are well-founded, their timing is open to debate. McKinsey’s research indicates that the United States has achieved only 18 percent of its automation potential (Manyika et al., 2017). Other research suggests that the rate of technological adoption remained relatively constant for the past 60 years, taking eight to twenty-eight years from commercial availability to 80 percent coverage (Bessen et al., 2019).

So, we need to make important distinctions between various stages of technology implementation to better understand the pace of impact on the workforce:

1. Feasibility: The technology works.
2. Commercial availability: The technology is reliable, economical, and available.
3. Implementation and assimilation: The technology is implemented, incorporated into the organization’s work processes, and delivers expected benefits.

The results from automation, including its impact on the workforce, are only achieved at the third stage. The inventor/producer of the technology is responsible for the first two stages. Once the technology is commercially available, the buyer is responsible for the third stage: implementation and full assimilation. This stage typically accounts for the longest time and requires a clear definition of the endeavor, including the following ingredients:

• Definition of objectives and expected benefits, with clear timeframes and measurable outcomes.
• Human resource initiatives to transition the workers affected by the technology and prepare those required for the implementation.
• Resources for implementation, including funding, skills, and technologies.
• Governance, defining the stakeholders required to steer the implementation to success and supported by appropriate management controls to navigate the implementation and assure achievement of goals.

In my personal experience and that of my colleagues at Prometheus, oversight of any of these ingredients leads to delays, partial implementation, missed objectives, and ultimate failure.

Covid-19, still in progress at the time of this writing, is drastically changing technology automation implementation dynamics.  A recent survey of 899 C-level executives and senior managers indicates the Pandemic has forced companies worldwide to accelerate the automation of their customer and supply-chain interactions and their internal operations by three to four years. To meet new demands from their customers, they have accelerated digitally enabled offerings by a shocking seven years.

Executives of these corporations surprised themselves with their organization’s ability to pull this off. Their mindsets on technology’s strategic importance appear to have changed radically during the crisis.  According to the survey, many of these solutions are temporary, but companies seem committed to making investments necessary to make them permanent.

History will tell how successful and enduring these changes will be and whether companies have learned how to improve their deployed technology automation solutions (McKinsey & Company, 2020).

Is the Worker under siege by technology automation?

For workers with low education performing predictable and repetitive tasks, the answer is YES. And, as we saw, this number will increase in the future. For these workers, the question is not “if” but “when.” Unless they prepare, they will need to resign themselves to low paying jobs when machines take their place. But for the majority, their displacement is not imminent as society has been historically slow to implement new inventions.

For workers with higher education, carrying out unpredictable tasks requiring high cognitive and motor skills, and perhaps creative and social intelligence, the answer is NO.  The good news for these workers is that technology offers a higher quality of work-life through less physical wear and tear and more challenging intellectual and social engagement.

Looking at society, we observe another dichotomy: while technology continues to improve our overall wealth and standard of living, it is also moving a growing segment of the workforce towards poverty. Thus far, the economy has grown faster than the rate of worker displacement by automation, assuring the possibility of full employment, albeit at lower wages. In the future, if economic growth falls behind the pace of worker displacement by technology, job prospects for these displaced workers will become hopeless.

As history teaches us, chronic unemployment and poverty lead to a breakdown of society and social unrest. As the power of organized labor dwindles in our gig economy, only political action will be able to address this problem.

Questions for further exploration

This report, which attempted to assess technology automation from the workforce perspective, leaves several questions to explore:

1. For business:  How should organizational leaders manage machines and workers in concert with one another, seeking to increase the value added of their employees?
2. For technologist: Can technology be the savior by enhancing jobs and sustaining quality employment, and not just displace jobs?
3. For educators:  Do we need to reinvent education to prepare new generations for a different workplace where work is increasingly done by machines? How do we help adults who find themselves displaced in the middle of their careers?
4. For workers:  What can workers do to insulate themselves from the negative impact of technology?  How can workers prepare to take advantage of new technologies to enhance their careers?
5. For society:  Should society prepare for the scenario in which further automation of physical and cognitive tasks fails to produce enough high quality jobs?

References

Acemoglu and Manera. (2020, March 18). Does the U.S. tax code favor automation? Retrieved from Brookins: https://www.brookings.edu/bpea-articles/does-the-u-s-tax-code-favor-automation/

Autor and Salomons. (2018, March). Is automation labor-displacing? Productivity growth, employment, and the labor share. Retrieved from Brookings Papers on Economic Activity: https://www.brookings.edu/wp-content/uploads/2018/03/1_autorsalomons.pdf

Bessen et al. (2019, February). Automatic Reaction – What Happens to Workers at Firms that Atomate? Retrieved from Boston University School of Law: https://scholarship.law.bu.edu/faculty_scholarship/584/

GAO. (2016). Report to the President.

GAO. (2019, March). WORKFORCE AUTOMATION, Better Data Needed to Assess and Plan for Effects of Advanced Technologies on Jobs. Retrieved from https://www.gao.gov/assets/700/697366.pdf

How robots change the world. (2018, June). Retrieved from Oxford Economics: http://resources.oxfordeconomics.com/how-robots-change-the-world

Lawrence, R. (2015, July 21). The Growing Gap between Real Wages and Labor Productivity. Retrieved from Peterson Institute for International Economics: https://www.piie.com/blogs/realtime-economic-issues-watch/growing-gap-between-real-wages-and-labor-productivity

Lecher, C. (2019, April 25). How Amazon automatically tracks and fires warehouse workers for ‘productivity. Retrieved from theverge.com: https://www.theverge.com/2019/4/25/18516004/amazon-warehouse-fulfillment-centers-productivity-firing-terminations

Manyika et al. (2017, November 28). JOBS LOST, JOBS GAINED:: WORKFORCE TRANSITIONS IN A TIME OF AUTOMATIION. Retrieved from McKinsey & Company: https://www.mckinsey.com/featured-insights/future-of-work/jobs-lost-jobs-gained-what-the-future-of-work-will-mean-for-jobs-skills-and-wages

McKinsey & Company. (2020, October 5). how-covid-19-has-pushed-companies-over-the-technology-tipping-point-and-transformed-business-forever. Retrieved from mckinsey.com: https://www.mckinsey.com/business-functions/strategy-and-corporate-finance/our-insights/how-covid-19-has-pushed-companies-over-the-technology-tipping-point-and-transformed-business-forever

Monahan et al. (circa 2017). 3D Printing and the Future of the U.S. Economy. Retrieved from ATKearney: https://www.kearney.com/documents/20152/888957/3D+Printing+and+the+Future+of+the+US+Economy.pdf/7719fc50-50b9-6194-4c4c-c3de38e9a88c

O*NET Online. (n.d.). Retrieved from https://www.onetonline.org/find/descriptor/browse/Skills/

Osborne, F. a. (2013, September 17). THE FUTURE OF EMPLOYMENT: HOW SUSCEPTIBLE ARE JOBS FROM COMPUTERISATION? Retrieved from Oxford Martin School: https://www.oxfordmartin.ox.ac.uk/publications/the-future-of-employment/

PwC. (2018, December). The macroeconomic impact of artificial intelligence. Retrieved from https://www.pwc.co.uk/economic-services/assets/macroeconomic-impact-of-ai-technical-report-feb-18.pdf

PWC UK. (2018). Will robots really steal your jobs? Retrieved from https://www.pwc.com/hu/hu/kiadvanyok/assets/pdf/impact_of_automation_on_jobs.pdf

Schwab. (2015, December 12). The Fourth Industrial Revolution What it Means and How to Respond. Retrieved from Foreign Affairs: https://www.foreignaffairs.com/articles/2015-12-12/fourth-industrial-revolution

Smith and Anderson. (2017, October 4). Automation in Everyday Life. Retrieved from Pew Research Center: https://www.pewresearch.org/internet/2017/10/04/automation-in-everyday-life/

U.S. Bureau of Labor Statistics. (2016, August). Current Employment Statistics survey: 100 years of employment, hours, and earnings. Retrieved from https://www.bls.gov/opub/mlr/2016/article/current-employment-statistics-survey-100-years-of-employment-hours-and-earnings.htm

Wikipedia. (n.d.). Engels’ pause. Retrieved from Wikipedia: https://en.wikipedia.org/wiki/Engels%27_pause

Wikipedia. (n.d.). Luddite. Retrieved from https://en.wikipedia.org/wiki/Luddite

Wohlsen, M. (2014, June 16). A Rare Peek Inside Amazon’s Massive Wish-Fulfilling Machine. Retrieved from WIRED: https://www.wired.com/2014/06/inside-amazon-warehouse/

Wright, J. (2017, January 26). In the Automation Debate, Don’t Forget the Job Multiplier Effect. Retrieved from Emsi: https://www.economicmodeling.com/2017/01/26/automation-debate-job-multiplier-effect/