The Rise of Artificially Intelligent Agents: AI’s Growing Effect on the Economy, PART 2

For more on the implications of humans and machines sharing resources and contributing to the economy, please see "The Rise of Artificially Intelligent Agents: AI’s Growing Effect on the Economy, Part 1."

In Homo Deus, Israeli historian Yuval Noah Harari offers a provocative vision of the future. Advances in artificial intelligence and biotechnology will allow the wealthy elites to re-engineer their brains and bodies and “upgrade themselves into gods.” The rest of humanity, unable to afford the cyborg enhancements, will suffer economic irrelevance and become “the useless class.”1

While some may dismiss such a scenario as science fiction, economist Anton Korinek thinks business and government leaders ought to take AI’s transformative potential seriously. “Sometime this century, AI will surpass human levels of general intelligence,” says Korinek. “When that happens, AI-enhanced elites or autonomous machines will dominate the labor market.”

Korinek, who holds joint appointments at the University of Virginia’s Department of Economics and Darden School of Business, believes that the proliferation of artificially intelligent agents (AIAs) will amplify economic disparities to such an extent that, without corrective action, life for the vast majority of Homo sapiens will become unlivable.  

Korinek’s Framework

To illuminate AIAs’ growing impact, Korinek introduced a novel framework, which describes both humans and intelligent machines as goal-oriented entities that share some basic economic properties. Both entities absorb scarce resources, which they need to ensure their survival and to fill their maintenance needs. “Humans consume food and machines absorb electricity,” says Korinek, “but the basic economic function is the same.” In addition, both humans and machines contribute to the economy through their labor.2  

High-Tech Firms as AIAs

According to Korinek, tech firms like Google and Facebook are the most tangible manifestations of AIAs today. Equipped with proprietary algorithms and vast amounts of data, they are capable of doing what no human has ever done before. In the process, they accumulate growing amounts of wealth. To maintain and improve themselves, tech companies absorb increasing amounts of the economy’s resources. Take the construction of server farms, on which tech companies spent $20 billion in 2018 alone. 

“Intelligence confers power. So far, we’ve been the most intelligent beings on the planet, and we own and control all resources. However, as machines become more sophisticated and autonomous, they will be exerting more and more power.”
Professor Anton Korinek

Korinek’s framework demonstrates that AIAs are a rising force. “We can see that all those resources,” says Korinek, “tens of billions of dollars, are devoted to the advancement of the AIAs. All factors that are valuable for AIAs, like specialized labor, are rising in price. And economic growth in the AIA realm is much faster than in the human realm.” According to Korinek’s estimates, the AIA-economy has seen double-digit growth in the past decade, whereas the human economy has almost stagnated. 

Fierce Competition for the Scarce Resources

What will ordinary humans do once smart algorithms become super-smart? Korinek observes that as intelligent machines start outperforming humans in a growing number of areas, humans may end up competing with AIAs over scarce resources.

How will this play out? 

In a recent paper, Korinek and Joseph Stiglitz, Nobel laureate in economics, consider two scenarios. In one, the wealthiest humans will acquire the latest technology to enhance themselves and attain superhuman intelligence. “The enhanced humans,” says Korinek, “will effectively be considered as a separate species of AIAs.” In another scenario, superintelligent autonomous machines will separate themselves entirely from humanity.3 As Korinek puts it, “An economy of the machines, by the machines and for the machines is possible. Humans won’t be needed for it to function.” 

In both cases, ordinary Homo sapiens lose out. The enhanced humans and super-intelligent machines will become more productive than the unenhanced by orders of magnitude, which will lead to vast increases in income inequality. Eventually, super-intelligent entities will command a growing share of the scarce resources in the economy, pushing the vast majority of humans below their subsistence level in a modern-day version of the Malthusian catastrophe.4  

Economists Taking Action

In recent decades, technological advances helped fuel unprecedented wealth accumulation among a small elite, leaving behind large segments of the population. The rise of AIAs will only accelerate that trend.

“In principle,” says Korinek, “technological progress could make everybody better off, leading to unprecedented prosperity.” What’s needed, however, is sufficient political will, which seems lacking. To remedy that, Korinek has co-founded Economics for Inclusive Prosperity, a nonprofit that brings together academic economists committed to developing new policy ideas to counter technology-driven labor market disruptions and income disparities between the rich and poor.  

“We have to ensure,” says Korinek, “that humans prosper and obtain a fair share of the resources produced by our current human-machine economy. It’s one of the biggest challenges for humanity.” 
 

To read about Korinek’s policy approaches to countering rising inequality caused by technological progress, please see Labor in the Age of Automation and Artificial Intelligence.

• 1Yuval Noah Harari, Homo Deus: A Brief History of Tomorrow (New York: Harper, 2017), 322.
• 2Anton Korinek, “Labor in the Age of Automation and Artificial Intelligence,” ECONFIP, February 2019, https://econfip.org/policy-brief/labor-in-the-age-of-automation-and-artificial-intelligence/.
• 3Anton Korinek and Joseph E. Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment,” forthcoming in The Economics of Artificial Intelligence: An Agenda (NBER Conference Report), ed. Ajay K. Agrawal et al. (Chicago: University of Chicago Press, 2019).
• 4Ibid.