HUY NGUYEN

“Technology is a gift of God. After the gift of life, it is perhaps the greatest of God’s gifts. It is the mother of civilization, of arts and of sciences.” Freeman John Dyson – The mathematical physicist

ABSTRACT:

Artificial Intelligence (AI) promises to improve existing goods and services and, by enabling automation of many tasks, to increase significantly the efficiency with which they are produced. However, measured productivity growth has declined by half over the past decade, and real income has stagnated since the late 1990s for OCED countries and the United States. This paper considers the importance of complementary intangible investment in preparation for fully gaining the Artificial Intelligence benefit in term of increasing labor productivity growth. It does so by reviewing the main arguments from previous literature and by assessing them accordingly to the economic views and shows that the most impressive capabilities of AI, particularly those based on machine learning, have not yet diffused widely. More importantly, AI full effects will not be realized until waves of complementary innovations are developed and implemented. The required adjustment costs, organizational changes, and new skills can be modelled as a kind of intangible capital. A portion of the value of this intangible capital is already reflected in the market value of firms.

I. INTRODUCTION

About 70,000 years ago, organisms belonging to the species Homo sapiens started to form even more elaborate structures called cultures. The subsequent development of these human cultures is called history (Harari, 2014). However, despite many millennia of evolution, none of the events discussed so far has mattered very much, at least in comparison to something else – something that bent the curve of human history like nothing before or since (Morris, 2011) (Figure 1). Three important revolutions shaped the course of history: The Cognitive Revolution kick-started history about 70,000 years ago. The Agricultural Revolution sped it up about 12,000 years ago. Then, the Industrial Revolution, which was just over two hundred years ago, made a sudden change in our social development (Figure 2). It was the sum of several nearly simultaneous development in mechanical engineering, chemistry, metallurgy, and other disciplines. Now comes to the second machine age. Computers and other digital advances are doing for mental power – the ability to use our brains to understand and shape our environments – what the steam engine and its descendants did for muscle power.

Nowadays, the rapid advance in the field of “Artificial Intelligence”, which could be called as “the second wave of IT-based technology” (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018), has profound implications for the economy as well as society at large. Artificial intelligence has the potential to directly influence products and services and the tasks required to create these goods, with important implications for productivity, employment, and competition. The discussion around the recent patterns in aggregate productivity growth highlights a seeming contradiction. On the one hand, there are astonishing examples of potentially transformative new technologies that could significantly increase productivity and economic welfare. On the other hand, measured productivity growth over the past decade has slowed significantly. This deceleration is substantial, cutting productivity growth by half or more if its level in the decade preceding the slowdown. It is also widespread, having occurred throughout the OECD and, more recently, among many large emerging economies as well. (Syverson, 2017).

This paper will give evidence and explanations for the need for complementary intangible investment to gain AI’s benefits entirely in term of increasing labor productivity. Firstly, the numerous potential applications of AI will be presented, which is the reason why people have a forward-looking technological optimism about the AI future. Second, the paper provides a disappointing recent reality of labor productivity growth in some major countries. Finally, this “paradox” could be explained and addressed by the required complementary intangible capital investment and discusses a linkage between the complementary investment in AI and the labor productivity growth

II. HOW COULD AI INCREASE LABOR PRODUCTIVITY?

1. Investment trends in AI sector from 2009 to 2019

AI has existed for decades: processing voice to text or language translation; real-time traffic navigation; dynamically serving targeted advertisements based on personal data and browsing history and so on. We are supposed to see more widespread, scaled adoption of AI across sectors. Nowadays, more and more companies are investing in AI and managing the complicated process of adopting this new technology. For example, the four leaders in terms of the number of AI startups funded (the United Kingdom, Israel and Germany) attract 80% of the total amount of fund raised in this sector over the 2009-2019 period, representing $8.6 billion out of a total of approximately $10.8 billion in funds raised by AI startups (Figure 3). It is a very positive picture of European dynamism in term of AI investment. However, in comparison with the US market, Europe is not growing as fast as it could, the US remains the indisputable leader of AI startup dynamism. In 2018, the United States counted 70 exits for an overall investment of $4.5 billion and 510 transactions with average fundraising of around $10 million. Among the sector-specific applications of AI, healthcare and biotech witnessed a surge in European AI startups which are representing 13% of AI startups while entertainment, media, culture accounts for 9% followed by financial services (8%) and defense, security (4%) (Figure 4). An analysis of the International Journal of Computer Vision, the most cited European AI journal – between 2015 and 2019 highlights both the strength of AI investment in terms of R&D. As we could see from the Figure 5 And Figure 6, R&D in AI is monopolized at the global level under the tripolar structure composed of the US, China and the UK, which represent more than half of the institutions featured in the journal. At the European level, the UK, France and Germany based institutions represent two-thirds of the institutions featured in the journal.

A considerable investment flows into the AI sector with a witness of an AI development race among major countries. This phenomenon could raise a question about whether there is a hype of AI. If it is not the case, so what, why and how AI could be a potential role in firm-level and country-level development?

2. Two extraordinary AI’s skills

One way of looking at the last 150 years of economic progress is that it is driven by automation. The industrial revolution used steam and then electricity to automate many production processes. Relays, transistors, and semiconductors continued this trend. “Perhaps artificial intelligence is the next phase of this process rather than a discrete break” (Philippe Aghion, Benjamin F. Jones, Charles I. Jones , 2018). Indeed, historically, most computer programs were created by meticulously codifying human knowledge, step by step, mapping inputs to outputs as prescribed by the programmers. Computers have been replacing humans in carrying out a widening range of tasks – filing, bookkeeping, mortgage underwriting, installing windshields on automobile bodies and so on – the list becomes longer each year. However, there is still some type of work which could not be entirely computerized. “Computers have an advantage over humans in carrying out tasks that involve some kinds of information processing. Nevertheless, humans retain an advantage over computers in tasks requiring other kinds of information processing” (Frank Levy, Richard J. Murnane , 2005). “Computers are good at following rules but lousy at pattern recognition” (Brynjolfsson, Erik and Andrew McAfee, 2014). When expressed in computer code, some rule-related works could be replaced by computer thanks to adding algorithms. For example, the mortgage underwriter who decides whether a mortgage application should be approved, the full list of rules might include tests on the applicant’s liquid assets, the number of years with the current employer, and so on. An application that passed every test was approved. Each rule leads to a clean “yes/no” answer and by setting “yes” = 1 and “no” = 0, it is easy to imagine how a computer could be programmed to process mortgage application in this way. However, lie information processing tasks that cannot be boiled down to rules or algorithms. The basic daily example is driving. The driving task requires the human capacity for pattern recognition. The driver has to recognize what he or she is confronting. However, articulating this knowledge and embedding it in software for all but highly structured situations are enormously tricky tasks. Computers cannot easily substitute for humans in jobs like driving. Now, thanks to modern AI, our digital machines have escaped their narrow confines and started to demonstrate broad abilities in pattern recognition, perception, complex communication, and other domains that used to be exclusively human. “Machines that can complete cognitive tasks are even more important than machines that can accomplish physical ones” (Brynjolfsson, Erik and Andrew McAfee, 2014). We are going to see AI do more and more, and as this happens, costs and employment will go down while the outcomes will stay the same or even improve, and our lives will get better. Soon countless pieces of AI will be working on our behalf, often in the background. They will help us in areas ranging from trivial to substantive to life-changing. Trivial uses of AI include recognizing our friend’s faces in photos and recommending products. More substantive ones include automatically driving cars on the road, guiding robots in warehouses, and better matching jobs and job seekers. However, these remarkable advances pale against the life-changing potential of artificial intelligence.

3. Economically and Efficiently

AI is a wonder of modern science that has made a lot of things possible that were unthinkable before. Now thanks to AI, many things can be done more quickly and more effectively. AI has increased the efficiency and productivity of many things in the industry. For instance, the saline wastewater, which is widely generated by industry, can be used for a variety of purposes such as food processing, textile, leather tanning and petroleum industries. However, the composition of saline wastewater depends mainly on the product, supplies, number of units used in the process and the water sources. Thus, saline wastewater may contain high organic loads, oil, grease, suspended solids, phosphorus and nitrogen. These old systems, which are using biological treatment processes, have inadequate organic load removal. Applying AI model, based on the combination of artificial neural networks and genetic algorithms, can increase the organic load removal efficiency above 70% then improve wastewater treatment performance of complex saline industrial wastewaters (Alain R. Picos-Benítez; Juan D. López-Hincapié; Abraham U. Chávez-Ramírez; Adrián Rodríguez-García, 2017).

Another example is that a team from Google DeepMind recently trained an ensemble of neural networks to optimize power consumption in a data centre. By carefully tracking the data already collected from thousands of sensors tracking temperatures, electricity usage, and pump speeds, the system learned how to make adjustments in the operating parameters. As a result, the AI was able to reduce the amount of energy used for cooling by 40% compared to the levels achieved by human experts. Overall, data centre electricity costs in the US are about $6 billion per year, including about $2 billion just for cooling (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018). There are many reasons for this. First, instead of replacing jobs, AI’s automation is far more likely to target specific tasks within a role – particularly repetitive ones we would consider to be”low-value” (David H. Autor, Anna M. Salomons , 2017). By fundamentally changing the types of jobs that are being done, AI allows humans to focus on more meaningful works, which could improve efficiency and productivity (David Autor, Anna Salomons, 2018). Seconds, AI could be embraced for the productive savings by using complex calculations, routine tasks and pattern recognition. With these extraordinary abilities, AI can minimize the number of errors and mistake during the production process, then, reduce the costs and improve the efficiency of the manufacture.

4. AI Improvement

AI or machine learning systems are also designed to improve over time. Indeed, what sets them apart from earlier technologies is that they are designed to improve themselves over time. Instead of requiring an investor or developer to codify, or code, each step of a process to be automated, a machine learning algorithm can discover on its own a function that connects a set of inputs X to a set of outputs Y as long as it is given a sufficiently large set of labelled examples mapping some of the inputs to outputs (Brynjolfsson, Erik, Andrew McAfee, 2017). The improvements reflect on only the discovery of new algorithms and techniques, particularly for deep neural networks, but also their complementarities with vastly more powerful computer hardware and the availability of much larger digital datasets that can be used to train the systems (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018). More and more digital data is collected as a byproduct of digitizing operations, customer interactions, communications and other aspects of our lives, providing fodder for more and better machine learning application.

5. The spillover effect and Innovation

A similar application of AI could be implemented in a variety of commercial and industrial activities. For instance, manufacturing accounts for about $2.2 trillion of value-added each year. Manufacturing companies like GE are already using AI to forecast product demand, future customer maintenance needs, and analyze performance data coming from sensors on their capital equipment. Recent work on training deep neural network models to perceive objects and achieve sensorimotor control at the same time have yielded robots that can perform a variety of hand-eye coordination tasks. (Levine, Finn, Darrell, and Abbeel, 2016). (Liu, Gupta, Abbeel and Levine, 2017) trained robots to perform several household chores, like sweeping and pouring almonds into a pan, using a technique called imitation learning. In this approach, the robot learns to perform a task using a raw video demonstration of what it needs to do. These techniques will surely be essential for automating manufacturing processes in the future. The results suggest that AI may soon improve productivity in household production tasks as well (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018), which in 2010 were worth as much as $2.5 trillion in nonmarket value-added (Bridgman, Dugan, Lal, Osborne, and Villones, 2012).

Moreover, if we think of AI as a type of capital, precisely a type of intangible capital (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018), (M. O’Mahony, M. Vecchi, 2009) find that the spillover effect existence is belonging to an intangible-intensive industry by using data on five large OECD economies between 1988 and 1997. To be more specific, the paper finds that the firms operating in most R&D and skill-intensive sectors have from 2-5% higher productivity growth. Similarly, (A. Elnasri, K.J. Fox, 2017) study the case of intangible investments in Australia between 1993-2013. The authors also find that private intangible investments have a general positive TFP effect in Australia, interpreted as a spillover effect.

Last but not least, AI can spur a variety of complementary innovations. For instance, machine learning of AI has transformed the abilities of machines to perform many primary types of perception that enable a broader set of application (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018). The significant advances in AI have not been in the form of the “general problem solver” approaches; instead, recent advances in AI are by, and large innovations that require a significant level of human planning and that apply to a relatively narrow domain of problem-solving. Therefore, AI is an area where we might focus on the impact of innovation (improved performance) and diffusion (more widespread application) in terms of job displacement versus job enhancement. Consider machine vision – the ability to see and recognize objects, to label them in photos, and to interpret video streams. As error rates in identifying pedestrians improve from one per 30 frames to about one per 30 million frames, self-driving cars become increasingly feasible (Brynjolfsson, Erik, Andrew McAfee, 2017). (Iain M. Cockburn, Rebecca Henderson, Scott Stern, 2018) gives some quantitative empirical evidence on AI effect on innovation by estimating the evolution of different areas AI in terms of scientific and technical outputs of AI researchers as measured by the publication of papers and patens from 1990 through 2015. Together, these preliminary findings provide that the innovation indicators are rapidly developing while AI application is being applied in many sectors.

III. THE DISAPPOINTING LABOUR PRODUCTIVITY PERFORMANCE

Although the giant AI’s benefits discussed above hold great potential, there is little sign that they have yet affected aggregate productivity statistic. “Labor productivity growth rate in a board swath of developed economies fell in the mid-2000s and have stayed low since then.” (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018). For example, labor productivity growth in the OCED area remains weak and well below the pre-crisis rate. Since 2010, annual growth in labor productivity has slowed to 0.9% about half the rate recorded in the 2000-2005 pre-crisis period (Figure 7). The post-crisis slowdown in productivity growth affects all major sectors but mainly manufacturing, where productivity growth rates remain well below last-decade’s rates in most countries (Figure 8). Indeed, in Australia, Israel, and the United Kingdom productivity gains in manufacturing have been negligible since 2010. In the services sector, the picture has been more varied (Figure 9). In Central and Eastern European OCED economies, for example, the catch-up process has helped sustain relatively robust growth, picking up actively in Poland and Slovenia in the most recent years. However, productivity growth remains weak in most other economies, indeed, sclerotic in some, such as Italy and Greece. Even in influential countries, such as Germany, Denmark and France, it remains weak. Wage growth has recovered in many countries but remains below pre-crisis rates in most countries (Figure 10). Growth in real wages, adjusted for inflation (using the consumer price index), has improved almost across the board in recent years compared with the early recovery period but remains below pre-crisis rates in two-thirds of OECD countries. The United States is experiencing the same scenario of a slowdown in measured labor productivity growth. “From 2005 through 2015(Q3), labor productivity growth has averaged 1.3% per year. This is down from a trajectory of 2.8% average annual growth sustained over 1995-2004” (Syverson, 2017). However, these slowdowns do not appear to reflect the effects of the Great Recession only. “In major advanced economies, productivity growth was slowing prior to the Great Recession.” (Gilbert Cette, John G. Fernald, Benoit Mojon, 2016). Both capital deepening and total factor productivity (TFP) growth lead to labor productivity growth, and both seem to be playing a role in the slowdown (Andrews, Criscuolo, Gal , 2016). Disappointing technological progress can be tied to each of these components. TFP directly reflects such progress. Capital deepening is indirectly influenced by technological change because firms’ investment decisions respond to improvements in the capital’s current or expected marginal product.

IV. THE NEED OF COMPLEMENTARY INTANGIBLE INVESTMENT IN AI

1. AI is a General Purpose Technology (GPT)

The inconsistency between forward-looking technological optimism and backwards-looking disappointment gives us a hint that the full AI benefits have not been fully reaped. AI, which has the vast potential to be pervasive, to be improved upon over time, and to spawn complementary innovations, is one of the prominent candidates that embody the characteristics of general-purpose technologies (GPTs). However, “A GPT does not deliver productivity gains immediately upon arrival” (Jovanovic, Boyan, Peter L.Rousseau, 2005)The technology can be present and developed enough to allow some notion of its transformative effects even though it is not affecting current productivity levels in any discernible way. “AI will bring a positive productivity shock to most economic sectors. However, AI capital will need complementary investments in intangible capital, such as complementary investment in firm-specific human capital and organizational structures” (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018). This is precisely the state that the economy may be in now. GPTs can at one moment both be present and yet not affect current productivity growth if there is a need to build a sufficiently large stock of the new capital, or if complementary types of capital, both tangible and intangible, need to be identified, produced, and put in place to harness the GPT’s productivity benefits fully. (David, 1989) notes a similar phenomenon in the diffusion of electrification. At least half of US manufacturing establishments remained unelectrified until 1919, about 30 years after the shift to polyphase alternating current began. Initially, adoption was driven by simple cost savings in providing motive power. The most significant benefits came later when complementary innovations were made. Managers began to fundamentally re-organize work by replacing factories’ centralized power source and giving every individual machine its own electric motor. This enabled much more flexibility in the location of equipment and made possible active assembly lines of material flow.

2. The benefit of complementary intangible investment in AI

Consider the benefit of complementary investment in intangible capital when applying GPTs such as IT or AI in firms, (Brynjolfsson, Erik, Lorin Hitt, 2003) found that while small productivity benefits were associated with firms’ IT investment when one-year differences were considered, the benefits grew substantially as longer differences were examined, peaking after about seven years. They attributed his pattern to the need for complementary changes in business processes. If the firm applies the new technologies to its manufactory process, but there is no considerable adjustment to match the firm’s human capital to the new structure of production, the GPTs will not have any noticeable effect on firm’s productivity. “As computers become cheaper and more powerful, the business value of computers is limited less by computational capability and more by the ability of managers to invent new processes, procedures and organizational structures that leverage this capability” (E. Brynjolfsson, L.M. Hitt, 2000) . For instance, when implementing large enterprise planning systems, firms almost always spend several times more money on business process redesign and training than on the direct costs of hardware and software. In fact, (Brynjolfsson, Erik, Lorin Hitt, 2000) also highlighted how investment in Information and Communication Technology (ICT), which is one of the GPTs candidates, needs even higher commitments to modern forms of firms’ organizational structure and to firm-specific human capital to be effective. The authors estimate that the ratio between ICT and complementary intangible investments is 1:9.

Furthermore, (Bresnahan, Timothy, Erik Brynjolfsson, and Lorin Hitt, 2002) find evidence of three-way complementarities between IT, human capital, and organizational changes in the investment decisions and productivity levels. (Brynjolfsson, Erik, Lorin Hitt, Shinkyu Yang, 2002) show each dollar of IT capital stock is correlated with about $10 of market value. They interpret this as evidence of substantial IT-related intangible assets and show that firms that combine IT investment with a specific set of organizational practices are not just more productive: they also have disproportionately higher market values than firms that invest in only one or the other. “It is plausible that AI-associated intangibles could be a comparable or greater magnitude” (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018). This pattern in the data is consistent with a long stream of research on the importance of organizational and even cultural change when making GPTs investment such as IT, ICT or AI and technology investments more generally (Henderson, Rebecca, 2006) (Orlikowski, 1996)

However, such changes take substantial time and resources, contributing to organizational inertia. Firms are complex systems that require an extensive web of complementary assets to allow the GPT to transform the system entirely. Firms that are attempting transformation often must reevaluate and reconfigure not only their internal processes but often their supply and distribution chains as well (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018). For example, considering the retail sector, (Micheal D. Smith, Joseph Bailey, Erik Brynjolfsson, 1999) show that the difficulties incumbent retailers had in adapting their business processes to take full advantage of the internet and electronic commerce. Many complementary were required. The sector as a whole required the build-out of the entire distribution infrastructure and employee’s training. Customers had to be “retrained”. None of this could happen quickly.

Another example is the case of self-driving cars. Consider what happens to the current pools of vehicle production and vehicle operation workers when autonomous vehicles are introduced. Employment on the production side will initially increase to handle R&D, AI development, and new vehicle engineering. Moreover, learning curve issues could well imply lower productivity in manufacturing these vehicles during the early years (Steven D. Levitt, John A. List and Chad Syverson, 2013). Thus, labor input in the short run can actually increase, rather than decrease, for the same amount of vehicle production. These changes can take time, but managers and entrepreneurs will direct invention in ways that economize on the most expensive inputs (Acemoglu D. , Restrepo P., 2017). According to LeChatelier’s principle, elasticities will, therefore, tend to be greater in the long run than in the short run as quasi-fixed factors adjust (P. Milgrom, J. Roberts , 1996) .

3. The Predicted Labor Productivity Indicator

While implementing AI initially depresses labor productivity due to increasing the employment in R&D, the slow productivity growth today does not rule out faster productivity growth in the future. (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018) used the data of US, productivity indices from 1948 to 2016 and ran the regression to test whether the past productivity growth rates are the good predictors of future productivity growth. As it turns out, the data shows that it would have been hard to predict the decrease in productivity growth in the early 1970s or foresee the beneficial impact of information technology (IT) in the 1990s. The regressions in Table 1 allow for autocorrelation in error terms across years (1 lag). Table 2 shows the results which cluster the standard errors by decade. In both cases, the R² of these regressions is low, and the previous decade’s productivity growth does not have statistically discernable predictive power over the next decade’s growth. Although the intercept in the regression is significantly different from zero, the coefficient on the previous period’s growth is not statistically significant. The lack of explanatory power of past productivity growth is also apparent in the scatterplots (Figure 11)

Instead of relying only on past productivity statistics to predict productivity growth, we should consider the technological and innovation environment we expect to see shortly (Erik Brynjolfsson, Daniel Rock, Chad Syverson, 2018). Brynjolfsson et al. (2018) used data comparing the US labor productivity between a period after portable power technologies had been invented and were starting to be placed into production (1890 – 1940) and a period which IT technologies were implemented (1970 – now). The authors see that labor productivity during the portable power era shared remarkably similar patterns with the current series (Figure 10). In both eras, there was an initial period of roughly a quarter-century of relatively slow productivity growth. The productivity growth slowdown we have experienced after 2004 also has a parallel in the historical data, a slowdown from 1924 to 1932. As can be seen in the figure, and instructive to the point of whether a new wave of AI and associated technologies could re-accelerate productivity growth at the end of the portable power ear rose again, averaging 2.7 per cent per year between 1933 and 1940.

V. CONCLUSION

There is a bounty of AI benefits which can be implemented in a variety of our life activities. Thanks to two essential and extraordinary skills gaining in AI, which are perception and cognition, AI now can replace more and more human jobs which can help the production input cost such as employment wage or production material go down while the outcome is improved significantly, thus, increasing the labor productivity. Moreover, the spillover effect is another factor that brings AI plays an important role the country-level development. The AI application is not only be implemented in one sector but also a wide range of sectors in the national economy, which enables complementary innovations that could multiply their impact. However, the productivity growth has slowed down recently and what gains there have been are unevenly distributed, leaving many people with stagnating incomes, declining metrics of health and well-being. This gloomy scenario could suggest that the breakthroughs of AI technologies already demonstrated are not yet affecting much of the economy.

By surveying the literature at the country, industry and firm level, this paper found evidence of the increasing importance of business intangibles in explaining labor productivity growth dynamics. Moreover, according to the results in the surveyed papers, to fully reap benefits of investment in Artificial Intelligence (AI), complementary investments in business intangibles are also essential. It points to organizational complements such as new business processes, new employee’s skills and new organizational and industry structures as a significant driver of the contribution of AI. These complementary investments may be as much as an order of magnitude larger than then investments in the AI itself. However, both the AI investments and the complementary changes are costly, hard to measure. They take time to implement, and this can, at least initially, depress productivity as it is currently measured.

Realizing the benefits of AI is far from automatic. It will require effort and entrepreneurship to develop the needed complements, and adaptability at the individual, organizational, and societal levels to undertake the associated restructuring. Theory predicts that the winners will be those with the lowest adjustment costs and that put as many of the right complements in place as possible. This is partly a matter of good fortune, but with the right roadmap, it is also something for which they, and all of us, can prepare.

3. Economically and Efficiently

4. AI Improvement

5. The spillover effect and Innovation

III. THE DISAPPOINTING LABOUR PRODUCTIVITY PERFORMANCE

IV. THE NEED OF COMPLEMENTARY INTANGIBLE INVESTMENT IN AI

1. AI is a General Purpose Technology (GPT)

2. The benefit of complementary intangible investment in AI

3. The Predicted Labor Productivity Indicator

V. CONCLUSION

3. Economically and Efficiently

4. AI Improvement

5. The spillover effect and Innovation

III. THE DISAPPOINTING LABOUR PRODUCTIVITY PERFORMANCE

IV. THE NEED OF COMPLEMENTARY INTANGIBLE INVESTMENT IN AI

1. AI is a General Purpose Technology (GPT)

2. The benefit of complementary intangible investment in AI

3. The Predicted Labor Productivity Indicator

V. CONCLUSION

References

A. Elnasri, K.J. Fox. (2017). The Contribution of Research and Innovation to Productivity. Journal of Productivity Analysis, 47, 291-308.

Acemoglu D., Restrepo P. (2017). The race between machine and man: Implications of technology for growth, factor shares and employment. (N. B. Research, Ed.) 22252.

Alain R. Picos-Benítez; Juan D. López-Hincapié; Abraham U. Chávez-Ramírez; Adrián Rodríguez-García. (2017). Artificial intelligence based model for optimization of COD removal efficiency of an up-flow anaerobic sludge blanket reactor in the saline wastewater treatment. Water Science & Technology, 75(6), 1351-1361.

Andrews, Criscuolo, Gal . (2016). The Best versus the Rest: The Global Productivity Slowdown, Divergence across Firms and the Role of Public Policy. OECD Productivity Working Papers.

Bresnahan, Timothy, Erik Brynjolfsson, and Lorin Hitt. (2002). Information Technology, Workplace Organization, and the Demand for Skilled Labor: Firm-Level Evidence. Quarterly Journal of Economics, 117(1), 339-376.

Bridgman, Dugan, Lal, Osborne, and Villones. (2012). Accounting for Household Production in the National Accounts 1965 – 2010. Survey of Current Business, 92(5), 23-36.

Brynjolfsson, Erik and Andrew McAfee. (2011). Race Against the Machine. Digital Frontier.

Brynjolfsson, Erik and Andrew McAfee. (2014). The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies. WW Norton & Company.

Brynjolfsson, Erik, Lorin Hitt. (2003). Computing Productivity: Firm-level Evidence. Review of Economics and Statistics, 85(4), 793-808.

Brynjolfsson, Erik, Andrew McAfee. (2017). What’s Driving the Machine Learning Explosion? Harvard Business Review, 18, 3-11.

Brynjolfsson, Erik, Lorin Hitt. (2000). Beyond Computation: Information Technology Organizational Transformation and Business Performance. Journal of Economic Perspectives, 14(4), 23-48.

Brynjolfsson, Erik, Lorin Hitt, Shinkyu Yang. (2002). Intangible Assets: Computer and Organization Capital. Brookings Papers on Economic Activity, 2002(1).

David Autor, Anna Salomons. (2018). Is Automation Labor-Displacing? Productivity Growth, Employment, and The Labor Share. NBER Working Paper Series No.24871.

David H. Autor, Anna M. Salomons. (2017). Robocalypse Now – Does Productivity Growth Threaten Employment? European Central Bank Sintra Forum Conference Paper.

David, P. (1989). Computer and Dynamo: The Modern Productivity Paradox in A Not-Too Distant Mirror. The Warwick Economics Research Paper Series.

E. Brynjolfsson, L.M. Hitt. (2000). Beyond Computation: Information Technology, Organizational Transformation and Business Performance. Journal of Economic Perspectives, 14, 23-48.

Erik Brynjolfsson, Daniel Rock, Chad Syverson. (2018). Artificial Intelligence and The Modern Productivity Paradox: A Clash of Expectations and Statistics. NBER Chapters, in The Economics of Artificial Intelligence: An Agenda, 23-57.

Frank Levy, Richard J. Murnane . (2005). The New Division of Labor: How Computers Are Creating the Next Job Market. Princeton University Press.

Gilbert Cette, John G. Fernald, Benoit Mojon. (2016). The Pre-Great Recession Slowdown in Productivity. European Economic Review, 88, 3-20.

Harari, Y. N. (2014). Sapiens: A Brief History of Humankind. Harper.

Henderson, Rebecca. (2006). The Innovator’s Dilemma as a Problem of Organizational Competence. Journal of Product Innovation Management, 23, 5-11.

Iain M. Cockburn, Rebecca Henderson, Scott Stern. (2018). The Impact of Artificial Intelligence on Innovation: An Exploratory Analysis. NBER Chapter, in: The Economics of Artificial Intelligence: An Agenda, 115-146.

Jovanovic, Boyan, Peter L.Rousseau. (2005). General Purpose Technologies. Handbook of Economic Growth, 1B, 1181-1224.

Levine, Finn, Darrell, and Abbeel. (2016). End-to-end Traning of Deep Visuomotor Policies. Journal of Machine Learning Research, 17(39), 1-40.

Liu, Gupta, Abbeel and Levine. (2017). Imitation from Observation: Learning to Imitate Behaviors from Raw Video via Context Translation.

M. O’Mahony, M. Vecchi. (2009). R&D, Knowledge Spillovers and Company Productivity Performance. Research Policy, 38, 35-44.

Micheal D. Smith, Joseph Bailey, Erik Brynjolfsson. (1999). Understanding Digital Markets: Review and Assessment . MIT Press.

Morris, I. (2011). Why The West Rules — For Now: The Patterns of History, and What They Reveal About the Future . Picador .

Orlikowski, W. J. (1996). Improvising Organizational Transformation Over Time: A Situated Change Perspective. Information Systems Research, 7(1), 63-92.

P. Milgrom, J. Roberts . (1996). The LeChatelier Principle . American Economic Review , 173-179.

Philippe Aghion, Benjamin F. Jones, Charles I. Jones . (2018). Artificial Intelligence and Economic Growth. National Bureau of Economic Research in The Economics of Artificial Intelligence, 237-282.

Roth, F. (2019). Intangible Capital and Labour Productivity Growth: A Review of the Literature. Hamburg Discussion Papers in International Economics, No.4.

Steven D. Levitt, John A. List and Chad Syverson. (2013). Toward an Understanding of Learning by Doing: Evidence from an Automobile Plant. Journal of Political Economy, 121(4), 643-681.

Syverson, C. (2017). Challenges to Mismeasurement Explanations for the US Productivity Slowdown. Journal of Economic Perspectives, 31(2), 165-186.

Tables and Figures

Figure 1: Human Social Development Index

Source: Brynjolfsson, Erik and Andrew McAfee. (2014). The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies. WW Norton & Company.

Figure 2: Human Social Development Index After Steam Engine Was Introduced

Source: Brynjolfsson, Erik and Andrew McAfee. (2014). The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies. WW Norton & Company.