The Economics of Artificial Intelligence in the Workplace

Abstract 

A new wave of innovation, Artificial Intelligence, has evolved in the workplace. It is projected that Artificial Intelligence will produce profound effects on the society since it promises probable benefits (not forgetting risks) in areas such as science, law, business, defense, and education among others. It is also project that Artificial Intelligence will most likely affect the distribution of income and employment in the society. This paper argues that the use of Artificial Intelligence will reduce significantly the need for people to toil at their workplaces. It is also noted that some individuals fear that Artificial Intelligence will displace a large number of workers from their jobs because of automation. However, since most people would rather use their time for leisure activities, this paper urges the society to greet the consequences of Artificial Intelligence in work-eliminating activities with much enthusiasm. This paper discusses the economic impact of artificial intelligence in the workplace. 

Key words: Artificial Intelligence, Workplace, Employment, Unemployment

Artificial Intelligence (AI) and related computer science developments have seen the rise of dramatic classes of machines, which are capable of performing tasks that need perception, judgment, and reasoning that only humans were previously capable of doing (Nilsson, 2000). Will such inventions decrease the requirement for human toil and result in high levels of unemployment? Two opposing views have emerged in response to this question. For instance, some people argue that AI is similar to other forms of technologies, which have revolutionized the workplace through increased production such as control operations research, mechanical engineering, and others (Kolbjørnsrud, Amico, & Thomas, 2003). It is proposed, therefore, that AI could lead to an expansion of the economy as the previous technologies have done, which will cause an increase in the rates of the creation of employment. According to this perception, at worst, there is a likelihood that the types of jobs could shift, but not an overall reduction in the numbers of jobs created within the economy (Kolbjørnsrud, Amico, and Thomas, 2003). However, it is notable that conservative appraisals of the real ability of AI in the job market underpin such an argument.

There are others who concede to a rather compelling hypothesis concerning AI, one that sets AI apart from technologies that have been used to save on the costs of labor (Nilsson, 2000). In simple terms, this hypothesis considers that AI is capable of doing just about anything that humans can do (Nilsson, 2000). Certainly, it is noteworthy that AI is yet to attain the levels of humans in performance, but most scientists in the field believe that it might equal and surpass human intelligence in the future (Lin, Abney, and Bekey, 2011). Therefore, the main conclusion of this perspective is that as much as AI might create more employment, AI machines could still do such work, and remove the necessity of having more humans employed. Of course, it is needful to understand that the fact that most of the jobs could be automated does not necessarily imply that they will. According to Lin, Abney, and Bekey (2011), different factors underpin automation such as social, political, and economic. Economists consider the relative cost at specified levels of both quantity and quality of either having humans or machines at the workplace as the central parameter of the debate. This perspective identifies AI as being different from the rest of the technologies of labor saving since it is relatively inexpensive, and has the potential or reducing in the future (Lin, Abney and Bekey, 2011). 

The purpose of this paper, therefore, is to examine the probable economic effect of the US of AI. The paper concludes that AI usage does not give a chance for a realization of massive reductions in human labor amounts needed for the production of goods and services for global consumption. While considering that some understandable reasons have caused some people to fear the outcome of adoption of AI and related technologies, the paper adopts a convincing tone for people to consider AI as a blessing. The work considers arguments from both sides of the divide before slowly emphasizing the positive effects of AI and other computer related technologies. It synthesizes appropriate literature, which points to the fact that the use of new technologies, from both a historical and technical perspective, have not resulted in a reduction in the numbers of workers nor a stagnation of the economy. The mere fact that a well-performing economy means that a large portion of the citizenry has meaningful employment and better standards of living is a reason to herald the power of AI in transforming the economies of the world.

The Dwindling Requirement for Human Toil

Before starting a more technical argument of the economic benefits related to AI, there is a need to review a few generalized statements that are in existence regarding the likely consequences of automation. First, it is needful to consider the arguments for automation, which support the idea that AI will not cause unemployment. According to James Albus, a prominent robotics researcher, there is no evidence, historical and contemporary, that rapid growth in productivity results in job losses (Albus, 1983). He notes that the trend is in fact, quite the converse since firms that use the most efficient techniques of production prosper and hire more laborers. Such firms then expand their markets, which allow them to diversify into new lines of production. A similar argument is related to the fact that the case of unemployment is a major problem in the developing nations than it is for those that have developed (Nilsson, 2000). The fact that automation is not as pervasive in the developing world than it is in the industrialized one means that automation alone might not be the principle source of unemployment. It is notable that even while automation could make it possible for a business to perform each task with the least numbers of people, there exist a great many needs for satisfaction. For instance, Albus still argues that there is no limitation to the amounts of work and that more work could always be created (Albus, 1983). Albus further posited that work is easy to generate and that the volumes of work to be done will always surpass the amounts of laborers who can do it. It means, therefore, the problem is not to find as much work as it would be possible for both machines and humans to do, but to come up with mechanisms that could ensure that the wealth that the AI technology generates is distributed among the individuals that need it (Nilsson, 2000). 

Evidence indicates that several industries have resorted to the use of rapid automation of their activities without realizing a reduction in the overall levels of unemployment (Nilsson, 2000). For example, according to Earnest (1982) (cited in Nilsson, 2000), the banking industry of the US explains this spectacle. He notes that an increased productivity within the industry because of automation was accompanied by more demand for banking services, which caused a rapid rise in employment (approximately 50 percent) from 1970 to 1980. However, it is noted that a majority of the jobs in the banking industry transformed to involve a type of knowledge requirement, which is a specific characteristic of AI. Even while the rate of automation would be rapid, the process of changing all the factories in a given economy into fully automatic ones will demand a significant amount of labor (Kurweil et al., 2000). Those that opposed the view that AI is capable of doing anything posit that a large number of tasks exist, which could not be automated fully. For instance, most people suppose that it might prove quite difficult or unreasonable to automate such services such as primary school teaching, child care, marriage counseling among others (Kurzweil et al., 2000). They further consider that machines have the little ability to generate excellent pieces of music, literary works as well as other forms of artistic works. In fact, there are some who project that the future economy will be grounded on specialized types of creative and human-oriented services in a similar way the contemporary economy is founded on a wide array of services (Kurzweil et al., 2000). 

Conversely, people that oppose AI and the idea that it will create more job opportunities have several reasons for their justification. For instance, they propose that the levels of unemployment in most of the developed nations, which use the highest levels of advanced technologies, have remained static for some time (Maney, 2016). They further note that the troughs of the employment of such nations grow bigger with each business cycle. It is noted that as much as most people lose their jobs during times of economic recession, there is a significant number of former employees that would blame their plight on the effects of AI and related technologies (Maney, 2016). In fact, some economists support the idea that the modern economic world is already in a phase when large-scale unemployment occasioned by rapid automation is inevitable (Nelsson, 2000). One of the most notable examples is Wassily Leontief, a Nobel-prize winning economist, who once argued that the world was starting to initiate a gradual process that would see most people displaced from their jobs within the next three to four decades, which would create a big problem of unemployment (Nilsson, 2000). He likened the trends during his time to a century ago when horses were phased out by the invention of machinery and argued that what had happened to the horses was likely to happen to the population.

Another argument proposed against the rapid adoption of AI and related technologies because of its negative effects on the labor market is the tendency of the AI technologies to automate more white-collar jobs (Trend Magazine, 2017). It is reported that a majority of Americans work in information processing jobs, which includes categories such as managerial functions like making decisions, supervision, fact-gathering, coordination, communication and reporting. Associated with this category of employment are the subordinate functions such as paper-handling responsibilities of clerical officers. Literature reports a general fear by those who oppose the full automation of the economy since they perceive that AI is capable of doing such jobs (Trends Magazine, 2017). There are those who still consider that most of the jobs that AI will create (if such a prospect materializes) will only be low-skill jobs (Trend Magazine, 2017). For example, one study highlighted that most of the new forms of employment would not be those regarded as belonging to the high-technology occupations, nor will the utilization of modern technologies in the existing jobs require heavy upgrading of the workforce. The study also pointed out that the growth in low-technology jobs will most likely outstrip the expansion of the high-technology ones. Therefore, there exist many different arguments on both sides of the debate, which is why is imperative to adopt a more technical perspective in addressing the economics of artificial intelligence in the workplace. 

The Technical Perspective

Economists, from whom this argument seeks clarification, differ in their perspectives on the role of AI in the workplace. In fact, Wassily Leontief (already described) and Herbert Simon appeared to one the opposite sides of the divide. For instance, Simon appeared to appeal the law of comparative advantage, which according to him, was a standard rebuttal of the theory of economics to the idea that mechanization results in unemployment. For instance, Simon states that the law indicated that both machines and people could be employed regardless of their contributions to the levels of productivity of organizations (Nilsson, 2000). He further posited that through adjustment in the relative prices of labor as well as capital, it would appear that labor would be hired in processes in which it will be considered the most productive. However, he concedes to the fact that the law of comparative advantage fails to settle all fundamental issues. He particularly notes that as much as the law indicates that all labor would be hired in equilibrium at some wage, regardless of how the machines would become efficient, it fails to predict what that wage would then be. Therefore, it means that according to Simon, there is no real guarantee in the law of comparative advantage that there will be no drop in the real wages; that they would remain on top of the level of subsistence (Nilsson, 2000). The effects of the progress in the levels of technology can be further analyzed in a model economy. Simon noted that a change in the labor saving technology in a model economy would result in a decrease in the average time of labor required to produce a single unit of output. He summed that as much as the rates of interest stagnated, an advancement in the levels of technology has only the chance of raising the real wages. He notes that nothing related to the present transformations in the levels of automation would create economic effects than the mechanization and industrial eras had (Nilsson, 2000). However, it is notable that more current studies on AI indicate the opposite of this prediction. 

According to Trends Magazine (2017), AI has led to the transformation of the economy at a faster rate, approximately 3000 times, than did the Industrial Revolution. Proponents of the piece of a technology project that it will spur development of the economy globally through its transformative effect from one industry to another. It is argued that the US will be one of the biggest beneficiaries. For instance, calculations made by the World Bank indicate that six years after the end of the Great Recession at the middle of 2009, the GDP of the US increased by only 2.1 percent (World Bank, 2017). It is indicated that as AI enabled companies and institutions to raise their levels of productivity, the US will have a real advantage of realizing a mean growth in its GDP of more than 3.5 percent at the onset of 2017 through the next two decades. 

Even while such predictions would be conservative, another study produced results that emphasized the critical importance of AI on the economies of 12 different countries, which make up more than 50 percent of the economy of the entire world. The study found that the use of AI and related technologies will raise the productivity of labor by 40 percent through the transformation of the manner in which work is done and the creation of collaboration among machines and people within the workplace (Trends Magazine, 2017). The researchers posited that the maturity of AI would mean a propulsion of the levels of economic development and a potential remedy to the stunted productivity levels as well as shortages of labor that have been experienced within the recent decades. In specificity, the study found that the use of AI would raise the productivity levels by 40 per cent and double the yearly rates of economic growth of most nations going forward to 2035 (Trends Magazine, 2017). 

Recent studies are also consistent with the older ones in addressing the fear that the use of AI will reduce the levels of employment for citizens of any nation. For instance, the modern inventions of smart robots and driverless vehicles will make the occupations in such fields completely obsolete and push many people out of meaningful employment (Maney, 2017 ). In addition, Bank of America predicted recently that businesses in the US would be able to save as much as $9 trillion in the costs of employment since AI will automate most of the knowledge jobs (Trends Magazine, 2017). She also added that healthcare providers and manufacturers would be able to save as much as $8 trillion because machines will take up human jobs (Trends Magazine). However, it ought to be understood that similar arguments have emerged at the advent of each major piece of the technological revolution. For example, a British Economist, David Ricardo, wrote during the Industrial Revolution in 1821 that the use of machines is often quite detrimental to the interests of human workers (Kolbjørnsrud, Amico, and Thomas, 2003). John Maynard Keynes, a century later in 1930, diagnosed a novel disease, which he called technological unemployment. According to Keynes, technological unemployment resulted from the discovery by individuals the means of labor use economization, which outruns the pace at which the economy found new ways of utilizing the labor force (Nilsson, 2000). 

However, literature is quick to refute such claims. For example, it is argued that each of the previous technological revolution eras has created more jobs than it has destroyed (Kolbjørnsrud, Amico, and Thomas, 2003). Trend Magazine (2017), that argues that 41 percent of 1900 workers in the US worked on farms initially, proposes a similar argument. However, the invention of tractors alongside other agricultural machinery resulted in the industry only employing 2 per cent of the workers in the US by 2000 (Maney, 2016). He noted that automation has further resulted in the replacement of scores of factory workers, and that software and computer has forced countless office workers out of their jobs. However, as the study notes, the labor force of the US stood at 159, 486,000 individuals as by November 2016 despite the disruptions to the labor force (Trends Magazine, 2017). The question, therefore, becomes how come most of the workers are able to find more jobs when it is hypothesized that new technologies kill jobs. 

The explanation to this scenario is quite simple. According to Kolbjørnsrud, Amico and Thomas (2003), those tasks that cannot be automated fully are only complemented by the automation. It should be noted that most of the industries have multifaceted sets of inputs comprising of capital and labor, brawn and brains, rote repetition and creativity, intuitive judgment and technical mastery, inspiration and perspiration, judicious discretion application and adherence to rules (Kolbjørnsrud, Amico, and Thomas, 2003). Each of the inputs has a critical role to play, which implies that an improvement done on one of them does not obviate the requirement for the other. Therefore, it is largely argued that improvements in terms of productivity in a single set of tasks are accompanied most likely by a rise in the economic value of the rest of the tasks (Kolbjørnsrud, Amico, and Thomas, 2003). It means, additionally, that while the routine elements of a single job might be automated through the use of technologies such as AI, workers might spend much of their time on elements of the job, which create more value for clients. In fact, some proponents argue that it is apparent that workers get better opportunities of employment when their jobs undergo some automation provided they learn how to use the new tools of technology (Trends Magazine, 2017). 

The concept highlighted in the previous paragraph can be illustrated by the example of the Automated Teller Machines that were considered to eliminate the need for most of the human tellers at the banks at their initial installment in the 70s (Kolbjørnsrud, Amico, and Thomas, 2003). In fact, the numbers of such machines increased from 100,000 in 1995 to 400, 000 in 2010 in the US, which according to the assumption of the negative effect of automation on the creation of jobs, would have caused a rapid decline in the numbers of human tellers. However, statistics indicate a reverse trend; the numbers of tellers in the US rose from 500,000 in 1980 to 550,000 in 2010. This, according to literature, only occurred because of the positive effects of automation associated with the machines. For instance, the automation meant that the banks were able to lower the costs of operation for their branches, which allowed them to establish more branches. Supportive data for this argument is that there was a rise by more than 40 percent in the numbers of bank branches in the US from 1988 to 2004 (Maney 2016). As much as the numbers of tellers in each branch dropped by approximately 33 percent within the same period, the high number of branches implied that, there was an overall high demand for the tellers in the nation. In addition, it is argued that while the teller machines took up most of the human jobs, the use of such modern jobs allowed the banks to develop relationship-banking services. The services created more jobs through letting the banks know their clients on personal levels and talking to them concerning other offerings from the banks such as investment vehicles, student and automobile loans, CDs, and credit cards. 

Therefore, it is quite clear that as much as those who oppose the use of AI could be justified in claiming that the technology has the potential of reducing the employment levels, it is also clear that their opinion is challenged. Modernity has one basic feature, the need to cater for large populations of people (Nilsson, 2000). Therefore, the use of AI is an effective tool in diversifying the economy, which is why literature is consistent in reporting that new technologies have been accompanied by an increase in the rates of employment. It is illogical, for example, to argue that the economy could be doing well when large portions of citizens are jobless (Maney, 2016). AI and other technologies, according to literature, prove to be drivers of economic development, which according to the maxim stated above, means more and better employment opportunities. In fact, the world had better embraced the use of AI technologies at the workplace since it reduces the need for human toil while improving their quality of life through the provision of more job opportunities for future generations. Issues of output maximization and cost minimization, therefore, are of central concern to employers, but they do not solely mean that more workers will be forced out of their jobs. 

In conclusion, as this paper has demonstrated, the world had better embraced AI since it is a major driver of economic growth. For example, while liberalists may argue that AI reduces the levels of employment within an economy, a conservative approach taken in this paper has demonstrated otherwise. The paper concedes to the fact that automation could reduce the numbers of jobs within a given industry, but challenges this argument by considering such scenarios are only short-lived. Therefore, an overall effect of automation is the creation of new jobs since such technologies lower the costs of labor (one of the most significant factors of production) and allows firms to diversify its production activities. The scare for massive job losses, therefore, is overcome by economic growth benefits that new technologies provide such as the Industrial Revolution and AI. In fact, as this paper has evidenced, AI has faster rates of economic transformation (3000 times) than the Industrial Revolution movement. 

References

Albus, J. (1983). From Washington: The robot revolution: an interview with James Albus.  Communications of the ACM ,  26 (3), 179-180.

Kolbjørnsrud, V., Amico, R., & Thomas, R. J. (2003). The promise of artificial intelligence. AI magazine . 3 (3), 7

Kurzweil, R., Richter, R., Kurzweil, R., & Schneider, M. L. (2000).  The age of intelligent machines  (Vol. 579). Cambridge: MIT press .

Lin, P., Abney, K., & Bekey, G. A. (2011).  Robot ethics: the ethical and social implications of robotics . MIT press .

Maney, K. (2016). How Artificial Intelligence And Robots Will Radically Transform The Economy. Retrieved May 22, 2017 from http://www.newsweek.com/2016/12/09/robot-economy-artificial-intelligence-jobs-happy-ending-526467.html

Nilsson, N. J. (2000). Artificial intelligence, employment, and income.  AI magazine ,  5 (2), 5.

Trend Magazine (2017). "Harnessing AI-Driven Growth - Trends Magazine".   Trends Magazine . N. p., 2017. Web. 22 May 2017.

World Bank (2017). GDP growth (annual %) Data.worldbank.org . Retrieved 22 May 2017, from http://data.worldbank.org/indicator/NY.GDP.MKTP.KD.ZG?locations=US