Positive and Negative Implications of AI in Today’s Workforce

Artificial intelligence: Will it deny us work or save us from work? Or something in between? With reference to relevant literature provide an account of the current and possible future dynamics of digital labour

 The term “Artificial Intelligence” (AI) was first coined by American computer scientist, John McCarthy, in 1955 as a way to describe the “the science and engineering of making intelligent machines” (McCarthy 1998). McCarthy’s (1998) concept of this futuristic study was based on the conjecture that every aspect of learning or any feature of intelligence in principle can be simulated by a machine as the ultimate effort to reach human-level intelligence. While, the more modern-day definition of this phenomenal change in technology focuses on AI being a sub-field of computer science that has the ability to imitate human intelligence such as visual perception, decision-making, speech recognition and the translation between languages (Oxford Dictionary 2018). Consequently, these transformations in technology opened up new opportunities for individuals, society and the economy as it provides unprecedented options for a prosperous future. However, despite AI’s intrinsic benefits, the pervasive force is largely attributed to the poor performance of labour markets across advanced economies as it has aggravated the debate about technological unemployment. This is further ascertained by the fact that AI’s acceleration of the automation of tasks that have long required human labour has now been replaced, and the augmentation of human skills through machines has the potential to cause a disruption in society’s current livelihood (Ford 2013). Ergo, this essay will discuss both the positive and negative implications of AI in today’s workforce and how this technological phenomenon can move towards a symbiosis of humans and machines.

Although there are many ongoing disagreements about the driving forces behind high unemployment rates, one of the most susceptible factors points to the rise of computerisation in labour markets as a possible explanation for low job rates. However, according to Petropoulos (2017), AI’s near-term effect is not mass unemployment but a gradual increase of occupational polarisation whereby a person moves from a mid-skilled job to a low-paid least skill job, which typically requires non-routine manual skills. This was further captured in Goos and Manning’s (2007) work titled, “Lousy and Lovely Jobs” as they posit the reasons for the growing employment in high-income cognitive jobs and low-income manual occupations. Firstly, problem-solving skills are becoming more productive due to the fall in the price of computing; hence, there is a substantial growth in employment that consists of cognitive tasks where skilled-labour has a comparative advantage. While low-wage jobs require unstructured conversation and extensive, unstructured physical movement which most computing programming cannot abide by. The relative distinction between AI and different job level skills is based on the trajectory of current automation trends, and this can be illustrated through the following: (1) Automation in low paying jobs are difficult to penetrate as these jobs require unstructured physical activity and unstructured social interaction. (2) Automation in high-skilled jobs (including the ones AI created) are difficult to penetrate as these require unstructured cognitive activity and unstructured social interaction. (3) Automation in middle-skilled jobs is relatively higher in contrast to the former two as this usually involves higher levels of structured physical activity and cognitive tasks. The key conclusion to these findings is that the implementation of technology into the subset of core occupational tasks that were previously performed by middle-skill employees has now caused substantial change and arguably the cause of a structural shift towards lower income jobs (Levy 2018). 

However, looking further into the future, a new wave of automation and more advent technological-learning techniques will emerge. In which, these intelligent machines will be increasingly capable of performing high-skilled and possibly non-routine manual tasks (Petropoulos 2017). Thus, the real question then becomes: Will AI displace or increase productivity in the labour market? In order to answer this question, the technological breakthrough in labour markets in the Industrial Revolution will firstly be examined. This example can be presented through the case of automobiles in the 1900s. Since the introduction of the motor-engineered vehicles, it has revolutionised the mode of transportation as any horse-related jobs were diminished. Though this caused many people to lose their jobs, this change has also opened up new industries, and many new jobs were created. Consequently, dominating the world economy as this led to an overall net positive impact on employment (Makridakis 2017). Furthermore, more reports that project similar patterns have emerged, and studies have shown that in the short run, the displacement effect may dominate. While in the long run, when society is fully adapted to the insemination of the automation culture, there will be a productivity effect and have a positive effect on employment (Petropoulos 2017). But to what extent is this approach reliable? According to the researchers from the Global Mckinsey Institute, the rate of disruption caused by the AI is estimated to be happening ten times faster in current society and at a scale of 300 times more than the Industrial Revolution, ergo, having approximately 3000 times the impact (Dobbs et al. 2015). Moreover, as the AI is continuously progressing to simulate the complexity of the human brain, the dynamics of these intelligent machines will have additional implications in the labour market.

The second resolution to this matter focuses more on assessing the risks in each of these occupations that are susceptible to automation in the next few decades. Researchers at the Oxford University published a study about the likelihood of computerisation for different jobs, indicating 12 out of around 700 occupations have a 99% chance of being automated in the near future (Frey and Osborne 2013). Though the number may seem insignificant due to its large context, according to the Bank of England it is estimated that 15 million jobs may be at risk by 2030 in the United Kingdom (Chu 2015). The occupations that are most inclined to automation all share a predictable pattern that illustrates the same repetitive activities, that have the possibility to be replicated through machine-learning algorithms. These highly intensive routine occupations that focus on easily-programable tasks such as— data entry keyers, new account clerks, assembly line workers, technicians are all particularly vulnerable to the replacement of changing technologies (Frey and Osborne 2013). While some of these occupations are completely eradicated, the demand for others is also reduced.  Due to the decline of occupational growth and the increase adaptation of new automation technology, the rise of inequalities of wages combined with those effects poses a constraint in the economic environment (Harris et al. 2018). This is so as the market imbalance and growth-stifling levels of inequality will cause the governmental role to be rectified to fit into a new technological era marketplace, and government intervention will be needed to address the economic imbalance. Although these studies support the displacement effect of automation, it is important to remember that AI is not a single technology, but rather, a collection of technologies that are applied to specific tasks. Furthermore, the impact of new technology on the aggregate labour market is not only dependent on the industry they operate in but also other spillover effects and adjustments in the economy (Makridakis 2017). Thus, it is difficult to predict AI’s driven automation immediate impact on the economy, as the uneven effect is uncertain due to the variant job scopes within the occupation itself.

Nonetheless, albeit it being more challenging to assess AI’s productivity effect on employment, recent innovations in cloud computing and big data storage analysis has helped the AI positively by improving efficiency in manufacturing industries (David 2015). Though many theories careen towards the displacement of productivity, AI is, in fact, a paradoxical phenomenon that also acts as a capital-labour hybrid. This is so as the driven automation offers the ability to amplify and transcend the current capacity of the capital and labour to fuel economic growth (Brynjolfsson and Syverson 2018). Moreover, as these technologies demonstrate progress, productivity improvements, increased efficiencies, safety and convenience in the labour market, the transformation of automation benefits become increasingly clear (McKinsey Global Institute 2017). According to research, the AI market is valued at USD $1.36 billion and is expected to grow at a compound annual growth rate (CAGR) of 52% during the forecasted period of 2017 to 2025 (Wood 2017). Furthermore, empirical evidence proved that machine automation produces products that are of higher standards and more efficiently, leading to better business performance overall (Wood 2017). In addition to AI’s productivity effect, another core comparative advantage of AI’s driven automation is the absence of human biases. The computerisation of cognitive tasks can be designed to complete a range of jobs through an impartial algorithmic solution that eliminates any potential human errors (Frey and Osborne 2013). Humans, unlike machines, manifest subjective elements such as flawed observations, favourable preferences, skewed data and even a necessity like sleeping poses negative constraints in their occupational performance (Horvitz and Heckermen 1986). Thus, it can be argued that the extension of automation in human labour can enhance labour productivity and obtain higher accuracy in the interpretation of data.

Next, five broad factors that have the ability to influence the pace extent of automation in the future workforce will be examined. The first factor, Technical Feasibility illustrates the idea that though technology has been automating human activities for centuries, it is still continuously improving as there is a time lag between a technology being demonstrated and a viable product being developed using that technology (Mckinsey Global Institute 2017). This can be seen through the example of aviation in the 1900s. Although aviation pioneers Orville and Wilbur Wright invented flying an aircraft in 1903, it was only 11 years later that the first commercial flight was launched and it took another 12 years for the true birth of commercial aviation to take place (Dobbs et al. 2015). And while new technological advances are applied in today’s society, innovations that are highly rampant such as self-driving cars and digital personal assistants are still under development as they are often imperfect (Kim et al. 2013). Hence, to make the automated solution perfect, a lot of effort by scientist and engineers are still needed, as the time to invent, integrate and adapt the technology into the specific activity is a long and tedious process. The next factor, The Cost of Developing and Deploying Solutions instigates that the cost of developing the automated technologies is severely affected by the time and place as the incurred cost from the Research and Development (R&D) phase needs to be recouped (Mckinsey Global Institute 2017). It takes large capital expenditure to develop and engineer the automated technologies as many companies also need to invest in the purchase of physical infrastructures such as tooling and laboratories. Thus, all these costs affect the business for where and when the automation is adopted as the cost of deploying and developing the technologies is ever-changing. 

Thirdly, Labour Market Dynamics are also important as the cost associated with labour activities are often affected by the complex dynamics of the labour market (Harris et al. 2018). Since the supply of labour is caused by many factors such as demographics and individual skills/talents, the supply, demand and wages all vary over time. Moreover, if there is an oversupply of workers and is significantly cheaper than automation, the relative costs of labour can be a significant argument against it (Mckinsey Global Institute 2017). However, looking at the labour dynamics in the future, the deceleration of the current labour force growth in advanced economies could result in a USD $5.4 trillion GDP shortfall by 2030 (Harris et al. 2018). Furthermore, as more middle-class employees are being phased out by automation trends in most advanced economies, it is estimated by researchers that they will face at least a decade of disruption as the new automation will begin to transform and revolutionise many industries and the global workforce (Harris et al. 2018). The fourth and most impactful factor in the future of the global economy is the Economic Benefits derived from automation. These driven automation technologies have the potential to generate business and economic incentives as it drives corporate decision-making skills and higher productivity levels (Brynjolfsson and Syverson 2018). As noted in the aforementioned paragraphs, the productivity effect of automation includes performance gains, increased output and productivity, improved safety and higher quality products. Ergo, making this benefit one of the most substantial progress in the global economic perspective. However, despite explicit automation benefits in the workforce, historical comparisons suggest that tackling massive job loss from automation- including jobs that are eradicated and not replaced in the future, will pose serious macroeconomic and social challenge for most advanced economies (Harris et al. 2018). Depending on each countries’ local market, the differences may either mitigate or exacerbate the disruption in the future.

 Lastly, Regulatory and Social Acceptance is one of the most commonplace barriers to the implementation of automation. The impact of automation in any given country will differ based on its distinct economic, social and governmental structures as each face a variety of different challenges based on their countries principles, labour forces and national jurisdiction (Harris et al. 2018). Firstly, the AI has to abide by all federal and regulatory policies as the innovated technology is liable to safety, and moral code conducts (Mckinsey Global Institute 2017). By enforcing the development of the AI in a safe and controlled environment, it decreases the chance of any potential accidents from occurring and any legal product liability if the technology malfunctions (Bloomfield 2018). Hence, the future of the AI is largely based on each countries’ governmental intervention as the regulations are imposed to manage market imbalance and to ensure the concern of public safety. Next, looking at the driven automation from a social perspective, there are many ethical concerns surrounding the morality of the AI that can be concluded. This is especially attributed to the fear of humans being replaced by the automated robots as there is lack of job security and financial well-being (Bloomfield 2018). With highly intense social and political pressures against automation on employment, this could trigger a social backlash that may cause governmental regulations to stop or slow down the pace of automation (Harris et al. 2018). In addition, some individuals may also face discomfort with the idea of automation due to their personal preference and thus making an abrogating judgment on the usage of these advent technologies (Schmidt et al. 1978).

In conclusion, the concept of this futuristic study and its ongoing debate about automation’s impact on mass unemployment will not only reshape national economies and future industries, but there will be a structural change in the entire nature of work. Though there are many perceived benefits of the AI high productivity levels in the workforce, which offers the best opportunities for reversing the slowdown in global output growth resulting from demographics, the offset of lower levels of labour force growth across advanced economies will require its productivity growth to be at least 54% higher than in 1995 in order to stabilise the economy (Harris et al. 2018). Furthermore, the high displacement effect of robots replacing human activities will also require a frame of a governmental policy response to mitigate the negative effect. However, by examining the positive and negative implications of the AI’s effect on the labour market, it is still difficult to predict the potential impact of automation in the labour market with absolute certainty as the complex and dynamic systems of the labour market are constantly evolving.

References

• Artificial Intelligence [Def. 1]. (n.d.). Oxford Dictionary Online. In Oxford. Retrieved October 25, 2018, from < https://en.oxforddictionaries.com/definition/artificial_intelligence>
• Brynjolfsson, E., Rock, D., & Syverson, C. (2018). Artificial intelligence and the modern productivity paradox: A clash of expectations and statistics. In the Economics of Artificial Intelligence: An Agenda. University of Chicago Press.
• Chu, B. (2015) The Independent. Robots put 15 million UK jobs at risk, warns Bank of England. Retrieved October 30, 2018, from
• <https://www.independent.co.uk/news/business/news/15-million-uk-jobs-at-risk-from-robots-warns-bank-of-england-a6732381.html>
• David, H. (2015). Why are there still so many jobs? The history and future of workplace automation. Journal of Economic Perspectives, 29(3), 3-30.
• Dobbs, R., J. Manyika and J. Woetzel (2015), The Four Global Forces Breaking all the Trends, London, San Francisco, Shanghai: McKinsey Global Institute.
• Ford, M. (2013). Could artificial intelligence create an unemployment crisis?. Communications of the ACM, 56(7), 37-39.
• Frey, C. B., & Osborne, M. A. (2013). The future of employment: how susceptible are jobs to computerisation?. Technological forecasting and social change, 114, 254-280.
• Goos, M. and A. Manning (2007), ‘Lousy and Lovely Jobs: the Rising Polarization of Work in Britain’, The Review of Economics and Statistics, 89(1): 118–33.
• Harris, K. Kimson, A and Schwedel, A. (2018). Bain and Company. Labor 2030: The Collision of Demographics, Automation and Inequality, Retrieved November 16, from <https://www.bain.com/insights/labor-2030-the-collision-of-demographics-automation-and-inequality/>
• Horvitz, E., & Heckerman, D. (1986). The inconsistent use of measures of certainty in artificial intelligence research. In Machine Intelligence and Pattern Recognition (Vol. 4, pp. 137-151). North-Holland.
• Levy, F. (2018). Computers and populism: artificial intelligence, jobs, and politics in the near term. Oxford Review of Economic Policy, 34(3), 393-417.
• Makridakis, S. (2017). The forthcoming Artificial Intelligence (AI) revolution: Its impact on society and firms. Futures, 90, 46-60.
• McCarthy, J. (1998). What is artificial intelligence?.
• Petropoulos, G. (2017) THE IMPACT OF ARTIFICIAL INTELLIGENCE ON EMPLOYMENT. Praise for Work in the Digital Age, 119.
• Wood, L. (2017). Business Wire. Global Artificial Intelligence Market 2017-2025 – Growth Trends, Key Players, Competitive Strategies and Forecasts Report – Research and Markets, Retrieved November 16, from <https://www.businesswire.com/news/home/20171130005513/en/Global-Artificial-Intelligence-Market-2017-2025—Growth>
• Mckinsey&Company. (2017). Mckinsey Global Institute. A FUTURE THAT WORKS: AUTOMATION, EMPLOYMENT, AND PRODUCTIVITY. Retrieved from <https://www.mckinsey.com/~/media/McKinsey/Featured%20Insights/Digital%20Disruption/Harnessing%20automation%20for%20a%20future%20that%20works/MGI-A-future-that-works_Full-report.ashx>
• Kim, J., Kim, H., Lakshmanan, K., & Rajkumar, R. R. (2013). Parallel scheduling for cyber-physical systems: Analysis and case study on a self-driving car. In Proceedings of the ACM/IEEE 4th International Conference on Cyber-Physical Systems (pp. 31-40). ACM.
• Bloomfield, B. P. (2018). The question of artificial intelligence: Philosophical and sociological perspectives. Routledge.
• Schmidt, C. F., Sridharan, N. S., & Goodson, J. L. (1978). The plan recognition problem: An intersection of psychology and artificial intelligence. Artificial Intelligence, 11(1-2), 45-83.