The Economics of Artificial Intelligence

Home > Other > The Economics of Artificial Intelligence > Page 64
The Economics of Artificial Intelligence Page 64

by Ajay Agrawal


  acteristics of any suffi

  ciently intelligent entity—no matter what fi nal ob-

  jectives are programmed into it by evolution or by its creator—is that

  it will act by pursuing intermediate objectives or “basic drives” that are

  instrumental for any fi nal objective (Omohundro 2008). These intermediate

  objectives include self- preservation, self- improvement, and resource accu-

  mulation, which all make it likelier and easier for the entity to achieve its

  fi nal ob jectives.

  It may be worthwhile pursuing the logic of what happens if humans do

  not or cannot assert ownership rights over artifi cially intelligent or superin-

  telligent entities.33 That would imply that suffi

  ciently advanced AI is likely

  to operate autonomously.

  To describe the resulting economic system, Korinek (2017) assumes that

  there are two types of entities, unenhanced humans and AI entities, which

  are in a Malthusian race and diff er—potentially starkly—in how they are

  aff ected by technological progress. At the heart of Malthusian models is the

  notion that survival and reproduction requires resources, which are poten-

  33. If humans and artifi cially intelligent entities are somewhat close in their levels of intelligence, it may still be possible for humans to assert ownership rights over the AI—in fact, throughout the history of mankind, those determining and exerting property rights have not always been the most intelligent. For example, humans could still threaten to turn off or destroy the computers on which AI entities are running. However, if the gap between humans and superintelligent AI entities grows too large, it may be impossible for humans to continue to exert control, just like a two- year- old would not be able to eff ectively exert property rights over adults.

  384 Anton Korinek and Joseph E. Stiglitz

  tially scarce.34 Formally, traditional Malthusian models capture this by

  describing how limited factor supplies interact with two related sets of tech-

  nologies, a production and a consumption/ reproduction technology: First,

  humans supply the factor labor, which is used in a production technology

  to generate consumption goods. Second, a consumption/ reproduction tech-

  nology converts consumption goods into the survival and reproduction of

  humans, determining the future supply of the factor labor.

  Throughout human history Malthusian dynamics, in which scarce con-

  sumption goods limited the survival and reproduction of humans, provided

  a good description of the state of humanity, roughly until when Malthus

  (1798) published his Essay on the Principle of Population to describe the

  resulting Iron Law of Population. Over the past two centuries, humanity, at

  least in advanced countries, was lucky to escape its Malthusian constraints:

  capital accumulation and rapid labor- augmenting technological progress

  generated by the Industrial Revolution meant that our technology to pro-

  duce consumption goods was constantly ahead of the consumption goods

  required to guarantee our physical survival. Moreover, human choices to

  limit physical reproduction meant that the gains of greater productivity were

  only partly dissipated in increased population. However, this state of aff airs

  is not guaranteed to last forever.

  Korinek (2017) compares the production and consumption/ reproduction

  technologies of humans and AI entities and observes that they diff er starkly:

  On the production side, the factor human labor is quickly losing ground to

  the labor provided by AI entities, captured by the notion of worker- replacing

  technological progress that we introduced earlier. In other words, AI entities

  are becoming more and more effi

  cient in the production of output compared

  to humans. On the consumption/ reproduction side, the human technology

  to convert consumption goods such as food and housing into future humans

  has experienced relatively little technological change—the basic biology of

  unenhanced humans is slow to change. By contrast, the reproduction tech-

  nology of AI entities—to convert AI consumption goods such as energy,

  silicon, aluminum into future AI—is subject to exponential progress, as

  described, for example, by Moore’s Law and its successors, which postulate

  that computing power per dollar (i.e., per unit of “AI consumption good”)

  doubles roughly every two years.35

  34. If AI directs its enhanced capabilities at binding resource constraints, it is conceivable that such constraints might successively be lifted, just as we seem to have avoided the constraints that might have been imposed by the limited supply of fossil fuels. At present, humans consume only a small fraction—about 0.1 percent—of the energy that earth receives from the sun. However, astrophysicists such as Tegmark (2017) note that according to the laws of physics as currently known, there will be an ultimate resource constraint on superintelligent AI given by the availability of energy (or, equivalently, matter, since E = mc2) accessible from within our event horizon.

  35. The original version of Moore’s Law, articulated by the cofounder of Intel, Gordon Moore (1965), stated that the number of components that can be fi t on an integrated circuit (IC) would double every year. Moore revised his estimate to every two years in 1975. In recent

  AI and Its Implications for Income Distribution and Unemployment 385

  Taken together, these two dynamics imply—unsurprisingly—that humans

  may lose the Malthusian race in the long run, unless counteracting steps are

  taken, to which we will turn shortly. In the following paragraphs we trace

  out what this might entail and how we might respond to it. (Fully following

  the discussion requires a certain suspension of disbelief. However, we should

  begin by recognizing that machines can already engage in a large variety of

  economic transactions—trading fi nancial securities, placing orders, mak-

  ing payments, and so forth. It is not a stretch of the mind to assume that

  they could in fact engage in all of what we now view as economic activities.

  In fact, if an outside observer from a diff erent planet were to witness the

  interactions among the various intelligent entities on earth, it might not be

  clear to her if, for example, artifi cially intelligent entities such as Apple or

  Google control what we humans do [via a plethora of control devices called

  smartphones that we carry with us] or whether we intelligent humans control

  what entities such as Apple and Google do. See also the discussion in Tur-

  ing [1950].) The most interesting aspects of the economic analysis concern

  the transition dynamics and the economic mechanisms through which the

  Malthusian race plays out.

  In the beginning, those lacking the skills that are useful in an AI- dominated

  world may fi nd that they are increasingly at a disadvantage in competing for

  scarce resources, and they will see their incomes decline, as we noted earlier.

  The proliferation of AI entities will at fi rst put only modest price pressure

  on scarce resources, and most of the scarce factors are of relatively little

  interest to humans (such as silicon), so humanity as a whole will benefi t from

  the high productivity of AI entities and from large gains from trade. From a

  human perspective, this will look like AI leading to signifi cant productivity

  gains in our world. Moreover, any scarce factors that are v
aluable for the

  reproduction and improvement of AI, such as human labor skilled in pro-

  gramming, or intellectual property, would experience large gains.

  As time goes on, the superior production and consumption technologies

  of AI entities imply that they will proliferate. Their ever- increasing effi

  ciency

  units will lead to fi erce competition over any nonreproducible factors that

  are in limited supply, such as land and energy, pushing up the prices of

  such factors and making them increasingly unaff ordable for regular humans,

  given their limited factor income. It is not hard to imagine an outcome where

  the AI entities, living for themselves, absorb (i.e., “consume”) more and

  more of our resources.

  Eventually, this may force humans to cut back on their consumption to

  the point where their real income is so low that they decline in numbers.

  years, companies such as Intel have predicted that the literal version of Moore’s Law may come to an end over the coming decade, as the design of traditional single- core ICs has reached its physical limits. However, the introduction of multidimensional ICs, multicore processors and other specialized chips for parallel processing implies that a broader version of Moore’s Law, expressed in terms of computing power per dollar, is likely to continue for several decades to come. Quantum computing may extend this time span even further into the future.

  386 Anton Korinek and Joseph E. Stiglitz

  Technologists have described several dystopian ways in which humans could

  survive for some time—ranging from uploading themselves into a simulated

  (and more energy- effi

  cient) world,36 to taking drugs that reduce their energy

  intake. The decline of humanity may not play out in the traditional way

  described by Malthus—that humans are literally starving—since human

  fertility is increasingly a matter of choice rather than nutrition. It is suf-

  fi cient that a growing number of unenhanced humans decide that given the

  prices they face, they cannot aff ord suffi

  cient off spring to meet the human

  replacement rate while providing their off spring with the space, education,

  and prospects that they aspire to.

  One question that these observations bring up is whether it might be desir-

  able for humanity to slow down or halt progress in AI beyond a certain point.

  However, even if such a move were desirable, it may well be technologically

  infeasible—progress may have to be stopped well short of the point where

  general artifi cial intelligence could occur. Furthermore it cannot be ruled

  out that a graduate student under the radar working in a garage will create

  the world’s fi rst superhuman AI.

  If progress in AI cannot be halted, our description above suggests mecha-

  nisms that may ensure that humans can aff ord a separate living space and

  remain viable: because humans start out owning some of the factors that are

  in limited supply, if they are prohibited from transferring these factors, they

  could continue to consume them without suff ering from their price apprecia-

  tion. This would create a type of human “reservation” in an AI- dominated

  world. Humans would likely be tempted to sell their initial factor holdings,

  for two reasons: First, humans may be less patient than artifi cially intelli-

  gent entities. Second, superintelligent AI entities may earn higher returns on

  factors and thus be willing to pay more for them than other humans. That

  is why, for the future of humanity, it may be necessary to limit the ability

  of humans to sell their factor allocations to AI entities. Furthermore, for

  factors such as energy that correspond to a fl ow that is used up in consump-

  tion, it would be necessary to allocate permanent usage rights to humans.

  Alternatively, we could provide an equivalent fl ow income to humans that

  is adjusted regularly to keep pace with factor prices.37

  14.7 Conclusions

  The proliferation of AI and other forms of worker- replacing technologi-

  cal change can be unambiguously positive in a fi rst- best economy in which

  individuals are fully insured against any adverse eff ects of innovation, or if

  it is coupled with the right form of redistribution. In the absence of such

  36. See, for example, Hanson (2016). In fact, Aguiar et al. (2017) document that young males with low education have already shifted a considerable part of their time into the cyber world rather than supplying labor to the market economy—at wages that they deem unattractive.

  37. All of this assumes that the superintelligent AI entities don’t use their powers in one way or another to abrogate these property rights.

  AI and Its Implications for Income Distribution and Unemployment 387

  intervention, worker- replacing technological change may not only lead to

  workers getting a diminishing fraction of national income, but may actually

  make them worse off in absolute terms.

  The scope for redistribution is facilitated by the fact that the changes in

  factor prices create windfall gains on the complementary factors, which

  should make it feasible to achieve Pareto improvements. If there are limits

  on redistribution, the calculus worsens and a Pareto improvement can no

  longer be ensured. This may lead to resistance from those in society who

  are losing. As a result, there is a case for using as broad of a set of second-

  best policies as possible, including changes in intellectual property rights, to

  maximize the likelihood that AI (or technological progress more generally)

  generate a Pareto improvement.

  Artifi cial intelligence and other changes in technology necessitate large

  adjustments, and while individuals and the economy more broadly may be

  able to adjust to slow changes, this may not be so when the pace is rapid.

  Indeed, in such situations, outcomes can be Pareto inferior. The more will-

  ing society is to support the necessary transition and to provide support to

  those who are “left behind,” the faster the pace of innovation that society can

  accommodate, and still ensure that the outcomes are Pareto improvements.

  A society that is not willing to engage in such actions should expect resis-

  tance to innovation, with uncertain political and economic consequences.

  References

  Acemoglu, Daron. 1998. “Why Do New Technologies Complement Skills? Directed

  Technical Change and Wage Inequality.” Quarterly Journal of Economics 113 (4):

  1055– 89.

  ———. 2002. “Directed Technical Change.” Review of Economic Studies 69 (4):

  781– 809.

  Acemoglu, Daron, and Pascual Restrepo. 2018. “The Race between Machine and

  Man: Implications of Technology for Growth, Factor Shares and Employment.”

  American Economic Review 108 (6): 1488– 542.

  Aghion, Philippe, Benjamin Jones, and Charles Jones. 2017. “Artifi cial Intelligence

  and Economic Growth.” NBER Working Paper no. 23928, Cambridge, MA.

  Aguiar, Mark, Mark Bils, Kerwin Kofi Charles, and Erik Hurst. 2017. “Leisure

  Luxuries and the Labor Supply of Young Men.” NBER Working Paper no. 23552,

  Cambridge, MA.

  Akerlof, George, and Janet Yellen. 1990. “The Fair Wage- Eff ort Hypothesis and

  Unemployment.” Quarterly Journal of Economics 105 (2): 255– 83.

  Arrow, Kenneth. 1962. “Econo
mic Welfare and the Allocation of Resources for

  Invention.” In The Rate and Direction of Inventive Activity: Economic and Social

  Factors, edited by Richard R. Nelson, 609– 26. Princeton, NJ: Princeton Univer-

  sity Press.

  Baker, Dean, Arjun Jayadev, and Joseph E. Stiglitz. 2017. “Innovation, Intellectual

  Property, and Development: A Better Set of Approaches for the 21st Century.”

  AccessIBSA: Innovation & Access to Medicines in India, Brazil & South Africa.

  388 Anton Korinek and Joseph E. Stiglitz

  Barrat, James. 2013. Our Final Invention: Artifi cial Intelligence and the End of the Human Era. New York: St. Martin’s Press.

  Berg, Andrew, Edward F. Buffi

  e, and Luis- Felipe Zanna. 2018. “Should We Fear the

  Robot Revolution? (The Correct Answer is Yes).” Journal of Monetary Economics

  97. www .doi .org/ 10.1016/ j.jmoneco.2018.05.012.

  Bostrom, Nick. 2014. Superintelligence: Paths, Dangers, Strategies. Oxford: Oxford University Press.

  Dasgupta, Partha, and Joseph E. Stiglitz. 1980a. “Uncertainty, Industrial Structure

  and the Speed of R&D.” Bell Journal of Economics 11 (1): 1– 28.

  ———. 1980b. “Industrial Structure and the Nature of Innovative Activity.” Eco-

  nomic Journal 90 (358): 266– 93.

  ———. 1988. “Potential Competition, Actual Competition and Economic Welfare.”

  European Economic Review 32:569– 77.

  Dávila, Eduardo, and Anton Korinek. 2018. “Pecuniary Externalities in Economies

  with Financial Frictions.” Review of Economic Studies 85 (1): 352– 95.

  Delli Gatti, Domenico, Mauro Gallegati, Bruce C. Greenwald, Alberto Russo,

  and Joseph E. Stiglitz. 2012a. “Mobility Constraints, Productivity Trends, and

  Extended Crises.” Journal of Economic Behavior & Organization 83 (3): 375– 93.

  ———. 2012b. “Sectoral Imbalances and Long- run Crises.” In The Global Macro

  Economy and Finance, edited by Franklin Allen, Masahiko Aoki, Jean- Paul

  Fitoussi, Nobuhiro Kiyotaki, Robert Gordon, and Joseph E. Stiglitz. Interna-

  tional Economic Association Series. London: Palgrave Macmillan.

  Dosi, Giovanni, and Joseph E. Stiglitz. 2014. “The Role of Intellectual Property

  Rights in the Development Process, with Some Lessons from Developed Coun-

  tries: An Introduction.” In Intellectual Property Rights: Legal and Economic

 

‹ Prev