Shu-Heng Chen AI-ECON Research Center, Department of Economics, National Chengchi University, Taipei, Taiwan
This chapter provides an economic study of the Internet of Everything (IoE). The fundamental question pursued by us is the impact of IoE on economic theory, specifically, the rationality of individuals in economic theory or in the history of economic analysis. To this end we examine individuals in the era of conventional economic theory in contrast to individuals in the era of the IoE. This comparison motivates us to sketch the future of individuals in economic theory. Two decades ago the Nobel Laureate Richard Thaler proposed two possibilities on this branching point: homo economicus, as people are depicted in neoclassical economics, and homo sapiens, as they are articulated in behavioral economics. Without the awareness of IoE, Thaler predicted that the latter would be the trend. In this paper, we shall address his prediction in light of the advent of IoE, and address the following two possibilities, namely, trend reversal and trend sustaining.
Homo sapiens; Homo economicus; Internet of Everything (IoE); Unmanned markets; Walrasian auctioneer; Ubiquitous networking; Economics of attention
An earlier version of the chapter was presented at the AAAI, 2018 Spring Symposium on Artificial Intelligence for the Internet of Everything at Stanford University, March 26–28, 2018. The author is grateful to the organizer of the symposium, Professor William Lawless, specifically for the invitation generously extended. The author is also grateful for the support for this research received from the Ministry of Science and Technology (MOST) [grant number MOST 106-240-2410-H-004-006-MY2].
It was the best of times, it was the worst of times, it was the age of wisdom, it was the age of foolishness, it was the epoch of belief, it was the epoch of incredulity, it was the season of Light, it was the season of Darkness, it was the spring of hope, it was the winter of despair, we had everything before us, we had nothing before us, we were all going direct to Heaven, we were all going direct the other way …
(Charles Dickens, 1859, A Tale of Two Cities, Book the First, Chapter I.)
What would be an “effective characterization” of the current era with its great advances in artificial intelligence (AI)? Although we could refer to everything around us as “smart”1 and these smart things could now be further connected together as a smart universe, called the Internet of Everything (IoE), this universe seems to be having difficulty pulling itself out of the mud that has trapped the old universe for so long.2 There are alarms and alarmists everywhere in our “smart” surroundings. It seems that every single timely issue that has already troubled us for long may also be “upgraded” with the advent of modern AI. A long but nonexhaustive list includes human intelligence and civilization, economic instability, employment, income inequality, dishonesty, discrimination, populism, polarization, social exclusion, privacy and security, democracy, etc. Celebrities such as Elon Musk (Tesla Motors), Bill Gates (Microsoft), Steve Wozniak (Apple), and Stephen Hawking (1942–2018) have all voiced their alarms in one way or another.
Given this conundrum, maybe Charles Dickens’ (1812–1870) magnum opus, A Tale of Two Cities, and his leading message, as we quoted above at the beginning of the chapter, stands as the most powerful illumination of what the era of AI can mean for us. In this chapter, we shall embark on a research project on the potential economic implications of IoE. As we shall argue, this research project, if placed in the history of economic analysis is not new (Section 10.2); however, IoE can be regarded as the advancement in technology from ubiquitous computing to ubiquitous networking, and its impact is unprecedented and overwhelming. Therefore the framework required to address this issue will be more comprehensive and extensive than the ones we have used before. This revisit will bring in new challenges associated with new concepts for economists.
As the first attempt to address this issue, our strategy is twofold. At the macro level, we shall examine the possibility of the unmanned markets as a continuation of the earlier socialist calculation debate (Section 10.2), whereas at the micro level we shall examine the rationality of individuals in economic theory in light of the long-standing debate between homo economicus and homo sapiens (Section 10.3). It may be worthwhile noticing that the earlier socialist calculation debate did not explicitly involve the micro (individual) part, while assumptions regarding individuals were made somewhat opaquely. In this chapter, as we shall see, the concept of individuals as cyborgs or extended minds enable this debate to have a bottom-up version. In other words, the modern version of the socialist calculation debate needs to begin with the issue of the rationality of individuals.
The rest of this chapter is organized as follows. Section 10.2 provides a review of the socialist calculation debate. Section 10.3 first presents the idea of cyborgs, and then assesses the impact of IoE on the rationality of individuals in economic theory. Concluding remarks follow in Section 10.4.
We have been through various episodes of industrial revolutions or, if expressed in a less overwhelming way, technological innovations. They each have their “dark side,” echoing well with what Joseph Schumpeter (1883–1950) referred to as “creative destruction” (Schumpeter, 1942). Hence, from the viewpoint of the history of economic analysis, if we wish to sympathize with some of the critics, pessimists, or skeptics, we need to ask why this time is different.3
This time is different because, unlike the past “industrial revolutions” or “technological shocks,” the current AI revolution driven by the advance in information and communication technology (ICT) is fundamental in the Walrasian sense (Walras, 1874). Interestingly enough as Dickens worked on one side of the Channel on his “Tale” his French counterpart in the other city (Paris), Leon Walras (1834–1910), an economist, had just started his academic career and 15 years later published his own magnum opus, Elements of Pure Economics (Walras, 1874). While this book has constantly been viewed as a groundbreaking step in economic analysis (specifically, the use of mathematical analysis in economics), the society for which the mathematics in his book operated has generally been ignored by those working in modern ICT and data scientists. In his book, Walras introduced an ideal type of economic society in which buyers and sellers not only have full information about the prices in the market, but could also participate in market transactions in an unrestricted (frictionless) manner. Markets are characterized by a high degree of organization and transactions, such as auctions, are centralized, and the terms of trade are made known openly to all market participants who, in turn, are given the opportunity to react to the prevailing prices.
Since both Charles Dickens and Leon Walras had experienced the political and economic upheavals after the Industrial Revolution, their work, especially that of the latter, regardless of how distant it may be from us in time, could still be useful as we experience the digital revolution, which, undoubtedly, is more significant in its impacts than previous industrial revolutions.
For the purpose of this chapter the significance of Walras’ Elements of Pure Economics is twofold. First, it can be read as the first book in economics that portrays the economy in the era of ICT and big data, and, second, based on that portrait, the entire market can be unmanned, with all the trading media replaced by machines. Of course Walras did not mention anything explicitly about ICT in Elements, but the entirety of his thought experiments regarding general equilibrium analysis cannot be taken seriously without some proper assumptions related to the availability of ICT. In fact, the Walrasian auctioneer, his brilliant intellectual invention, can be interpreted as a platform (network of supercomputers) employing a large number of robot traders or software agents to perform cloud trading. In this sense, one may treat his perception of markets as IoE, which came to the mind of this genius way ahead of his time.
This idea of an unmanned market (the Walrasian auctioneer) was then further developed by socialist economists, notably, Oscar Lange (1904–1965) (Lange & Taylor, 1938) and Abba Lerner (1903–1982) (Lerner, 1944), becoming the foundation of the socialist economy and economic planning. Along this intellectual line, the idea of cybernetics, which originated from the earlier servo-mechanism (automation) and was popularized by Norbert Wiener (1894–1964), was introduced into economics by Oscar Lange, who also authored his posthumous publication Introduction to Economic Cybernetics (Lange, 1970). In the 1970s and 1980s, economic cybernetics was further pursued by Western economists under a new subject title, referred to as optimal control theory (Kendrick, 1981).
Despite its long burgeoning period of development, the idea of an unmanned market, that is, economic planning as a substitute for the market economy, has been constantly questioned and challenged, and constitutes what is known as the socialist calculation debate (Boettke, 2000). Ludwig von Mises (1881–1973) and Friedrich Hayek (1899–1992) led the charge against socialism; this debate, while alien to most computer scientists and AI researchers, can be considered to be the longest and the largest-scale debate on the possibility and desirability (consequences) of AI.4 Alongside this debate, Hayek published his pathbreaking book, The Road to Serfdom (Hayek, 1944), which is probably the earliest penetrating and thought-provoking warning against unlimited exposure to AI. Notice that Hayek did not object to the use of machines or chess games with robot players; he raised objections to the use of AI to institutionalize the market mechanism.
So, is this time different? Even though ICT in the last century was insufficient to make the Walrasian auctioneer “smart enough” to support a centralized unmanned market, the current ICT revolution is in a better position to exhibit that function. Therefore what concerns us this time is not the usual job-loss alarm, which does not make this time different, but rather the market-annihilation alarm, which does. Insofar as Hayek’s warning is valid, Helbing et al. (2017) can then be read as a refreshing message of Hayek (1944): instead of empowering the state over individuals, technology can now empower the platform over individuals. In any case, we could be under the shadow of Thomas Hobbes' (1588–1679) Leviathan (Hobbes, 1977).
Economics deals with the causes and consequences of human decision (choice) making in resource allocation subject to various physical constraints and desires. Mainstream economics has long established a standard representation for the rationality of this decision-making agent, which is normally known as homo economicus or the von Neumann-Morgenstern expected-utility maximizing agent (Von Neumann & Morgenstern, 1944). These agents can learn from experiences and hence can cope with the future without making systematic errors. They can also solve various trade-offs in their lives: consumption and saving, portfolio returns and risk, education and employment, work and leisure, diet and health, residential location and commute flexibility, numbers of kids, etc. Apart from the accidents, incidents and natural disasters that are beyond their control, their rationality implies that they have self-control to carry out the choices made above so that their lives can be as happy as possible, probably even better than what is archived in George Vaillant’s Aging Well (Vaillant, 2008). The chaos depicted in Charles Mackay’s classic Extraordinary Popular Delusions and the Madness of Crowds (MacKay, 1841) may never occur in their “real” lives.
Obviously not everyone enjoys this self-portrait,5 and efforts made to search for the alternatives contributed to the formation of a new kind of homo, known as homo sapiens, that has become the statue of behavioral economists. The 2017 Nobel Laureate in Economics, Richard Thaler, predicts that models of individuals in economic theory will be characterized by a paradigm shift from homo economicus to homo sapiens (Thaler, 2000). His prediction was made almost two decades ago, when the current ICT revolution was not yet fully developed. If humans can be empowered with intelligent wearable devices, bionic chips, and IoE, will the evolving cyborgs eventually reverse this trend and revive homo economicus?
It seems to be a good place to detour slightly to involve the term “cyborg” (short for cybernetic organisms) here when one is addressing the economic consequences of IoE. If placed in a proper historical context, IoE may be regarded as the further implementation of cyborg science. The term cyborg was coined in 1960 by Manfred Clynes and Nathan Kline (Clynes & Kline, 1960). As Manfred Clynes recalled in an interview, “I thought it would be good to have a new concept, a concept of persons who can free themselves from the constraints of the environment to the extent that they wished. And I coined this word cyborg” (Gray, 1995, p. 47). Is not the statement also the pursuit for those who are so enthusiastically devoted to IoE? Despite this being the case, IoE differs from the idea of cyborgs by placing its focus, not on individuals, but on societies. Hence the picture is not just one of a man “travelling” through different “spaces” freely in the sense of communication and control, but more about the novel prototypes of interaction and coordination that can emerge with the availability of these extending “spaces.”
The relationship between cyborg science and economics has been splendidly reviewed by the economic historian, Philip Mirowski. Mirowski (2002) suggested that ideas pertaining to machines can be regarded as a thread connecting a myriad of branches of economics developing side by side with neoclassical economics after World War II that include computational economics, artificial economies, autonomous agents, and experimental economics; hence, cyborg science is closely associated with the history of economic analysis. Earlier we discussed the socialist calculation debate and the idea of the unmanned market, already pointing to cybernetics. Our revisit here will enable us to more clearly see why IoE will inevitably get involved in the continuation of this debate, while in different forms and with different rhetoric, primarily because its “formal body” has already been placed in the very heart of economics.
Indeed, if one reflects upon the history of behavioral economics and is sufficiently open-minded to include those great novels that deal with the complexity (or perplexity) and uncertainty of human nature, the core issue remains human decision making under limited intelligence, cognitive capability or bounded rationality, and, very recently, emotional influences (Wälde & Moors, 2017). With the advancement in AI and the ubiquitous IoE, it is feasible now to enhance the human’s cognitive functionality with various cyborg-like extensions, using not just external (wearable) devices, such as smart phones or Google glasses, but also internal (implant) devices, such as the extended mind (Clark & Chalmers, 1998; Rosenfeld & Wong, 2017) and also the shared (networked) powerful “mind” (IoE). It then seems relevant to ask whether the increasingly enhanced cyborgs will become more like homo economicus.
In the following, we propose two contrasting possibilities in response to Thaler’s prediction in light of IoE, which will be distinguished by different names: one is called trend reversal (Section 10.3.2) and the other trend sustaining (Section 10.3.3). As for the former, Thaler’s prediction will be overthrown, whereas for the latter it will be fulfilled.
The main argument that IoE could support trend reversal lies in the possibility that IoE can facilitate human decision making or even automate it. However, that enhancement or automation has already existed for a long while. What is new is that, through IoE, this decision-supporting function can be further enhanced not just at the individual level (optimization), but further up to a social level (interaction and coordination) by means of ubiquitous networking.6 A frictionless economy, as well presented by the neoclassical economy, will then be just around the corner.7
To elaborate on the above point, we first recognize that various intelligence tools (robots, software agents) have been extensively applied to “learn” and to “know” humans using the big data collected from IoE. These intelligent artificial agents can discover the repertoire of each decision maker in terms of preferences, moral judgments, beliefs, risk preferences, and routines or habitual behavior; hence, not only can IoE predict what individuals will choose, but, moreover, in their best interests, what they ought to choose; furthermore, if authorized, they will execute the choices for them as their incarnations. Since machines have little difficulty in exercising self-control, they can execute the choice with little hesitation.8
If these artificial agents are further applied to automate human decisions to a rather extensive extent, instead of merely driving an automobile, they can “routinize” many nontrivial decisions for humans. In this way, through IoE, networks of human agents are essentially mapped to networks of their cyborg equivalents. The entire (dual) economy is then run with enhanced rationality and coordination; the coordination scale is much more extensive than the idea of a city having only driverless cars or programmed drivers (Sasaki & Flann, 2005). In this manner, humans through their incarnations will behave as if they are homo economicus. This in turn helps with the formation of a frictionless economy.
Now, let us leave the unmanned vehicles behind, and focus on unmanned markets. With the information availability supported by IoE, the market mechanism as manifested by its automated matching mechanism can match couples or crowds to form partners, teams, firms, or various organizations. Matching per se is a highly complex issue, but good matches can enhance human well-being dramatically.9 Thanks to IoE, when the whole life span of humans can fundamentally be copied (mapped) into a cyber space, which is “topologically equivalent” to our real world, its digital landscape will be very “machine-friendly” for the artificial agents who are designed and dispatched to this space to search for and explore all possible opportunities.
As an illustration, consider the following example concerning the labor market. A college student upon his/her graduation has his/her resume automatically generated, which is then automatically sent to a matching pool where vacancies are matched with talents, followed by the provision of the most favorable deals for both (employer/employee) sides. A sequence of information processing, preference discovering, and skill endorsing, skill matching wage bargaining problems is automatically handled by artificial agents (robots) in cyberspace. When this happens, a collection of artificial agents takes over the labor market, which was for a long time run by humans, and the labor market as we understand it in the conventional sphere is gone and becomes unmanned.
One side effect associated with the advancement in matching technology carried out by the artificial agents in the IoE space is the development of the customization-oriented economy. The advancement in matching technology also facilitates the emergence of new production models, such as peer production, sharing and a pro-social economy. We can add more to extend this list, but what will happen if something we say is not correct?
The above argument is built upon the key assumption that we have artificial agents smart enough to learn from humans, to “nudge” them,10 and even replace them. However, in each of these aggressive steps, there are not only promises, but also pitfalls, which, if outweighed, can together paralyze the design of autonomous agents as human incarnations. In what follows we shall provide a short, but not exhaustive, list of impediments. The three points that we shall raise below are all concerned with the economics of attention (Lanham, 2006).
First, IoE has endowed us with a degree of information richness to an unprecedented level. Now, within the IoE environment, information is constantly flowing around us. Some of our idle time, such as waiting for the bus, the traffic light, the elevator, or the waiter, is now recruited to processing the constant inflow of information. We then gradually lose these idle moments, and certainly cannot entertain them with our conventional styles of dazing, dreaming, pondering, reflecting, or run-away thinking. A neologism, the smartphone zombie, has been invented to describe this unparalleled phenomenon. While the effect of IoE on our brains and minds is unclear at this point, it may be fair to say that it has shaped or defined what thinking can mean for us. Now the conventional way of thinking by “staring at the wall” or offline thinking is being dramatically replaced by online thinking, i.e., thinking with constant interruptions. Comparing these two different styles of thinking is not that straightforward, and while the familiar dual-task or multiple-task psychological experiments may help shed some light on this comparative study, further work needs to be done (Damos, 1991). However, when decision makers are often required to make their decisions online and spontaneously, the deliberation efforts normally required by homo economicus may be severely curtailed.
Second, a related point is that our attention capacity is a scarce resource and the information richness brought by IoE can make attention scarcity even more strained. In the literature this tension is known as information overload (Sutcliffe & Weick, 2008). Information overload can adversely affect our decision-making capability, which has already been well studied in psychology and behavioral economics (Iyengar, 2010). In fact, Herbert Simon (1916–2001) had long noticed such a tension (Simon, 1971). Although he did not immediately exclude the possibility that the advances in ICT may compound the information overload problem rather than solve or mitigate it, he did provide the condition required for avoiding the appearance of the downside.
“Whether a computer will contribute to the solution of an information-overload problem or instead compound it depends on the distribution of its own attention among four classes of activities: listening, storing, thinking and speaking. A general design principle can be put as follows: An information processing subsystem (a computer or new organization unit) will reduce the net demand on the rest of the organization’s attention only if it absorbs more information previously received by others than it produces—that is, if it listens and thinks more than it speaks.” (Ibid, p. 42; italics added). Briefly put, IoE has to make us feel quieter rather than noisier.
So far there is no clear evidence to show that our IoE environment can satisfy the Simon condition, and it is not entirely implausible to say that it tends to speak more than it can effectively listen and think (Chen, Chie, & Tai, 2016; Chen & Venkatachalam, 2017). If so, while a myriad of interconnections and interactions provided by IoE allow us to surf over a huge space of opportunities, it also exposes us to a potentially large number of decision problems, each with many alternatives. The latter is notoriously known as the choice overload problem or the paradox of choice (Iyengar & Lepper, 2000; Schwartz, 2004). Time and attention allowed for each of these choice problems is, therefore, severely diluted. Under such circumstances, to facilitate decision making, the attention-lacking agents may rely more on their fast track of information processing (i.e., the reflexive system), and less on their slow or deliberate track, the reflective system (Kahneman, 2011). In addition to emotion and gut feeling, various fast and frugal heuristics, such as following the herd, choice reinforcement, or using rules of thumb, will play a more contributory role in decision making (Gigerenzer & Gaissmaier, 2011), which may again make decision makers more like homo sapiens instead of homo economicus.
Finally, if information overload and choice overload have driven decision makers to behave more like homo sapiens, then even though machine learning can effectively extract and learn the behavioral patterns of these decision makers, the artificial agents that have been built may be, at best, another homo sapiens, since what was learned by artificial agents is what they actually did, but not what they ought to do for the sake of their own best interest. If one employs these artificial agents as the incarnation of their human counterparts and automates the decisions for them, then the well-known GIGO (garbage in, garbage out) principle may be applied (Stephens-Davidowitz & Pabon, 2017), and the things that are in action are again homo sapiens and not homo economicus.
The above three cases, while not exhaustive, justify why Thaler’s prediction remains valid, and is independent of the IoE technology.
Unlike the usual articles on IoE, which deal with a specific application of IoE, this chapter provides a panoramic view of IoE in economic theory, specifically, across the history of economic analysis. Reviewing the history of economic analysis in the light of developments in mathematics, physics, and machines has become a unique approach in economics. In that direction, this chapter can be regarded as a continuation of the intellectual inquiry initiated by Philip Mirowski (Mirowski, 2002, 2007).
Our inquiry has been carried out at both the macro and micro level. At the macro level, we explored the possibility of the unmanned markets as the continuation of the socialist calculation debate, while in the context of IoE, and at the micro level, we address the rationality or irrationality of individuals in economics under IoE. The two inquiries are closely related because the unmanned markets are largely built upon the automated decisions for individuals. Hence, these automated decisions, to some extent, determine the performance of the unmanned markets.11 When automated decisions take place on a large scale, the individuals’ decisions are facilitated not just by pens and pencils, abacuses, decision-support systems, or high-performance computing, but also by ubiquitous networking (IoE). With this “extended (augmented) mind,” would these individuals (cyborgs) behave more like homo economicus? Or, alternatively, would the prediction of Richard Thaler be affected by the presence of IoE?
In this chapter, we addressed the plausible models of individuals in economic theory, when these individuals become part of the IoE economy. This issue seems to be particularly relevant since in neoclassical economics decision makers have long been treated as information-processing units. This notion was already formalized when decision sciences and economics were fully integrated in von Neumann and Morgenstern’s collaborative magnum opus, Theory of Games and Economic Behavior (Von Neumann & Morgenstern, 1944), and was further strengthened when statistics also adopted a decision-making ontology and a game-theoretic methodology (Savage, 1954). The later rational expectations revolution in the 1970s and 1980s had pushed the concept of the information-processing unit to unprecedentedly high ground, probably achieving the golden age of homo economicus. However, this portrait has long worried psychologists and behavioral economists, and there seems to be a turn to homo sapiens (Zou & Chen, 2018).
In this chapter we provided supporting arguments on both sides. We first use the idea of economics as a cyborg science (Mirowski, 2002) and enriched that idea with ubiquitous networking to form an “optimistic” expectation that the fading-away homo economicus will be recruited back to the center of the IoE economy. If this happens, the high time of behavioral economics will come to a stop, and the “mainstream” (neoclassical economics) will once again triumph. We then used Simon’s economic theory of attention to indicate that the reverse expectation can also be plausible. According to Simon (1971), attention is a scarce resource. Simon warned us that the persistent deficiency in attention may not mitigate the problem of information overload, and as long as information overload remains, humans need to cope with these cognitive burdens with various heuristics and hence they are not immune to various biases and errors.12
If homo sapiens remains the major species in IoE times, what could the ecology of the IoE economy look like? What else will happen if machines do not make men smarter? This question is probably more philosophical than scientific, and some judgment needs to be exercised for questions like this. Earlier, Bauerlein (2008) and Carr (2011) argued how “things” can make humans “dumber”; interestingly, in a reverse direction, O’Neil (2017) and Stephens-Davidowitz and Pabon (2017) also enabled us to see how vice versa humans can make “things” dumber. Altogether, an ecological cycle of these foolish reciprocities can be self-constructing. While this proposition could run the risk of being an over-exaggeration, this saying recurs through history. “The road to serfdom” as “paved” by Friedrich Hayek (Hayek, 1944) has not become a relic; that “road,” once in a while is still filled with “pilgrims”; recently the road has been renovated (Helbing et al., 2017).
3.143.9.223