AI, jobs and economic prosperity

April 12 2019
by Joshua Levine


I believe in the sanctity of our soul. I believe there are a lot of things about us that we don't understand. I believe there's a lot of love and compassion that is not explainable in terms of neural networks and computation algorithms. Kai-Fu Lee

After earning a Ph.D. in computer science from Carnegie Mellon, the Taiwan-born Lee went to work in high-level positions at Apple, Microsoft and Google. Today, he is the CEO of Chinese venture capital firm Sinovation Ventures and the author of AI Superpowers: China, Silicon Valley and the New World Order. For his recent appearance on 60 Minutes, he was dubbed 'the oracle of AI.' Reports about his television interview seized on his prediction that 40% of jobs in the world – not just for blue-collar work, but a lot of white-collar work – will become "displaceable" by technology: "The invention of the steam engine, the sewing machine, electricity, have all displaced jobs. And we've gotten over it. The challenge of AI is this 40 percent, whether it is 15 or 25 years, is coming faster than the previous revolutions."

Now, it's important to understand that the 40% Lee refers to are jobs he believes are capable of being displaced, not necessarily rendered extinct. The reality is that technology mostly augments workers, not replaces them. As MIT professor David Autor argues, technology can't really do the vast majority of jobs: "Tasks that cannot be substituted by computerization are generally complemented by it. Most work processes draw upon a multifaceted set of inputs: labor and capital; brains and brawn; creativity and rote repetition; technical mastery and intuitive judgment; perspiration and inspiration; adherence to rules and judicious application of discretion."

Autor maintains that using technology to automate some part of a job almost always makes the tasks that the machine cannot do more valuable, because with technology, the value of the entire job goes up. This argument is supported by Erik Brynjolfsson and Tom Mitchell, who write in Science about the workforce implications of machine learning: "Although parts of many jobs may be 'suitable for ML,' other tasks within these same jobs do not fit the criteria for ML well; hence, effects on employment are more complex than the simple replacement and substitution story emphasized by some. Although economic effects of ML are relatively limited today, and we are not facing the imminent 'end of work' as is sometimes proclaimed, the implications for the economy and the workforce going forward are profound."

The 451 Take

Automation often reduces jobs in a specific industry. It also compels new investments as a result of a gained competitive advantage and improved customer experience, among other benefits. Emerging technologies like machine learning can directly lead to innovative products and new kinds of employment, increasing opportunities for both higher profits and wages. 451 Research data on job security indicates that the current state of the US economy is potentially strong enough to absorb many of the job losses stemming from technological change – indeed, only 13% of those surveyed say they worry a great deal or quite a bit about someone in their household losing their job.

ML adoption

A 2018 study by 451 Research examined organizations' plans to adopt machine-learning initiatives over the next three years. The results revealed that only 36% of respondents did not plan to implement ML in this time frame. Conversely, 65% of respondents either have current ML deployments in place or have plans to implement ML in the next two to three years. When asked what are the most significant benefits their company realized or expects to realize from its use of machine learning, nearly half (49%) said gaining competitive advantage. This was followed closely by improved customer experience (44%).

Interestingly, lower costs (25%) came in fifth place, which indicates that the replacement of labor by technology is not among the highest priorities for companies adopting machine learning, an application of AI that provides systems the ability to automatically learn and improve from experience without being explicitly programmed.

Figure 1
Figure 1
451 Research
These and other findings by 451 Research tell us that organizations view AI/ML as mostly augmenting rather than replacing human capabilities, automating the more routine parts of a job and increasing the productivity and quality of workers. But this doesn't mean the transition toward a more automated society over the next decade or two will be without serious challenges for companies or workers.

Dramatic changes ahead for economy

The Industrial Revolution, as tumultuous as it was, resulted in creating far more jobs and netted a huge increase in productivity. The advent of mills meant that hundreds of pieces of fabric could be made with the same amount of time and resources it previously took to make one. Greater productivity caused the price of textiles to come down, and demand soared. The general population benefited through steady wages and access to things they previously couldn't afford.

Although automation decreases jobs in a specific industry, it also creates new investments as a result of cost savings and increased profits. Automation can lead to wage increases as a result of greater profits. Prices are then likely to increase, and innovation directly leads to new products and new job opportunities.

So, were the Luddites wrong? It's not as simple as that, according to 451 Research. The Luddites' worries were based on two key assumptions:

  • First, that there is a finite amount of work to do, and if machines do the work, there is less left for humans to do. This is an incorrect assumption because as innovation occurs, new opportunities are invariably created.

  • The second assumption is that machines do 'easy' work, and that the definition of 'easy' expands as IT progresses. This means the work beyond 'easy' requires greater brain power than a computer can deliver – creativity, insight, gut feeling, and multidisciplinary talent and skills. With artificial intelligence, this balance of brain power is rapidly shifting.

  • A century and a half ago, 50% of the US labor force worked in agriculture. Even as agriculture shifted to less than 2% of the labor force, the total pool of workers boomed, filling jobs in technology, manufacturing and services.

    In the coming years, workers will have to shift accordingly as demand for AI, IoT, blockchain and cybersecurity skills rapidly increase. To help bridge people to the new workplace, higher education institutions need greater urgency. According to the most current data from the US Department of Education's National Center for Education Statistics, US colleges and universities conferred 286,133 science, technology, engineering and math degrees in 2016. Just to meet the projected shortfall, colleges would need to boost STEM degree production to 1.2 million diplomas – an increase of 415%.

    Companies also have to take on greater responsibility to manage such a transition. "Companies need to consider how their technologies will impact society as a whole," according to 451 Research's Dr. Owen Rogers. "This isn't social altruism; this is common-sense economics. Companies need to support retraining and innovation if they want to survive in the long term."

    When asked about the impact of AI/ML and other automation technologies on their own jobs and careers, workers express decidedly mixed views, according to Pew Research. Asked how likely it was that their jobs will be replaced by robots or computers in their lifetimes, 77% of fast food workers said it was somewhat or very likely. Insurance claims processors (65%) and software engineers (53%) also held pessimistic views. Alternatively, nurses (20%) and teachers (36%) see a better future.

    The World Economic Forum, a nonprofit group based in Geneva, provides its own readout on the jobs landscape for 2022, for the top emerging and declining roles as follows:

  • The Top Emerging Roles include: Data analysts and scientists, AI and machine learning specialists, software and applications developers and analysts, big-data specialists, digital transformation specialists, new technology specialists and information technology services. The WEF sees opportunities for these and others increasing by 133 million in the next few years.

  • The Top Declining Roles include: Data entry clerks, accounting, bookkeeping and payroll clerks, administrative and executive secretaries, assembly and factory workers, customer service workers, accountants and auditors, and stock-keeping clerks. Up to 75 million people in these positions are predicted to lose their jobs, according to WEF.

  • 451 Research data on job security indicates that the current state of the US economy is potentially strong enough to absorb many of the job losses stemming from technological change. As the chart below shows, only 13% of respondents worry a great deal or quite a bit about someone in their household losing their job. At net +22, this is 5 points better than last December. The number of respondents who do not worry at all (35%) remains unchanged year-over-year.

    Figure 2
    Figure 2
    451 Research
    Contrary to wide belief, a growing body of evidence suggests that workers have everything to gain from welcoming the robots, writes the Wall Street Journal's Christopher Mims. "The more robots a country has, the higher its gross domestic product and, on average, the richer its citizens. On the other hand, a country that resists automation loses out not just on wealth creation but on new jobs as well."

    A new report from the Information Technology and Innovation Foundation, one of the world's leading science and technology think tanks, argues that the US is falling behind in the adoption of robots. Its new index compares the rate of adoption of industrial robots in manufacturing in different countries, while controlling for average wages of workers in those countries and industries. The ITIF found the US is adopting industrial robots well behind the 'expected' rate of adoption, compared with other rich countries.

    China, on the other hand, is adopting robots so much faster than everyone else that, within a decade, it could lead the world in use of robots, when controlling for wages. Overall, the US ranks seventh in the world in its ratio of robots to manufacturing workers, but that only translates to two industrial robots per 100 workers. In South Korea, there are seven.

    The trouble with greater implementation of robots and AI, however, is that it potentially will worsen the inequality already disrupting the US. As firms cluster around talent, and talent is in turn drawn to those firms, the result is a self-reinforcing trend toward ever-richer, ever-costlier metro areas that are economically dominant over the rest of the country.

    In a paper published earlier this year, Robots and Jobs: Evidence from US Labor Markets, Daron Acemoglu and Pascual Restrepo, economists at MIT and Boston University, demonstrate that the Midwest and sections of the South have far higher ratios of robots to population than other regions of the US. They calculate the job losses resulting from the addition of one robot in a 'commuting zone.' Their bottom line: "One more robot in a commuting zone reduces employment by about six workers."

    Overall, according to David Autor, employment is growing steadily, and its growth in terms of number of jobs has not been discernibly dented by technological progress. But the sum of wage payments to workers is growing more slowly than economic value added, so labor's share of the pie of net earnings is falling. This doesn't mean that wages are falling. It means that they are not growing in lockstep with the value added.

    Acemoglu and Restrepo worry that the robot-related dislocations in automated industries will harm, and thus inflame, the discontent of key voters and lawmakers, even as jobs are created elsewhere. In their January paper, Artificial Intelligence, Automation and Work, they write:

    "Last but not least, the development and adoption of productivity-enhancing AI technologies cannot be taken for granted. If we do not find a way of creating shared prosperity from the productivity gains generated by AI, there is a danger that the political reaction to these new technologies may slow down or even completely stop their adoption and development. This underscores the importance of studying the distributional implications of AI, the political economy reactions to it, and the design of new and improved institutions for creating more broadly shared gains from these new technologies."