Artificial Intelligence and Society: Philosophy of Fallibility
Part 2: Dystopian Outlook on AI

KOBAYASHI Keiichiro
Faculty Fellow, RIETI

A worldview that assumes that AI is something that is beyond human reason overturns the familiar idea that humans are the pinnacle of creation. If human intellectual activity is nothing more than part of the motion of the universe and if AI arrives as something that transcends humans, what will become of human society?

In his book Homo Deus (Harari, 2018), the sequel to Sapiens (Harari, 2016), Yuval Noah Harari, a historian, sounded an alarm by describing a dystopian vision of the world that will be realized through the radical development of AI and biotechnology. In that vision, the ruling class (the ultra-wealthy), which numerically represents only a tiny minority of the population, will raise their intellectual capabilities with the help of AI to a level beyond the understanding of the humans of today. The ruling class will also acquire eternal youth thanks to biotechnology and new drugs. Harari coined the name "Homo Deus" for the tiny minority of humans who will achieve artificial evolution through technological innovation. Many tasks will be performed by AI and mechanization, while the remaining majority of people, who are no different from the humans of today, will degenerate into a "useless class" that has no role to play in society. Harari predicts that Homo Deus will dominate the useless class and may eventually wipe them out, just as homo sapiens hunted mammoths and other large mammals to extinction and drove the Neanderthals to extinction. The humans of today may also be drowned in the wave of human-led evolution and disappear. Harari's vision of the future of mankind described in this book stirs deep concerns.

However, Harari's insight into human society that the advance of technology causes significant disruptions is not new. It is an old, familiar argument that the advance of technology and endless competition may destroy people's lives in a capitalist economy or a liberal economy. This argument is distinctive in that it implicitly assumes that the tendency of inequality between the strong and the weak is to widen and that it will continue forever, with the strong ultimately wiping out the weak (the principle of natural selection). The Time Machine published in 1895 by H.G. Wells, the pioneer of the science fiction genre, describes a human society 800,000 years in the future. It is the future to come if the inequality between the capitalist class and the working class that existed before Wells' eyes at the end of the 19th century continues to widen forever. On the earth of 800,000 years in the future, mankind has evolved into two separate species. Descendants of the capitalist class have evolved into a frail species with a moderate disposition and subhuman intellectual and physical capabilities (the Eloi), while descendants of the working class have turned into a ferocious, cannibal species who live underground and consume the Eloi. This is a vision of the future that would be realized by the "principle of natural selection" if the inequality that existed in the 19th century continued to widen forever. The evolutionary logic of natural selection has been incorporated into the argument that Harari makes in Homo Deus.

It should be kept in mind that just as the inequality that existed in the 19th century did not continue to widen forever, the new inequality of our time (which is being realized by AI and biotechnology) may not necessarily keep widening perpetually. In response to a dystopian argument like that, a counterargument that inequality will (may) stop widening and start to narrow sooner or later, based on the theory of "directed technological change", presented by Professor Daron Acemoglu, is a possibility.

Daron Acemoglu, a professor of economics at the Massachusetts Institute of Technology argues that industrial technological change is biased toward saving scarce factors of production and exploiting abundant factors of production. This is known as the theory of directed technological change (Acemoglu [2002]). For example, in the 19th century, labor-intensive technological change occurred in the United Kingdom, where labor was abundant while land was scarce. In the United States, where labor was scarce but land was abundant, capital-intensive technological change occurred. The advance of AI is expected to make human labor useless in various industries. If the useless class as referred to by Harari—that is, unskilled workers living on low wages—expands, technological progress geared toward employing those people as a factor of production will occur. That is what is predicted by the theory of directed technological change. If this prediction comes true, demand for human labor will grow, resulting in a wage increase, causing the inequality to narrow.

After the extreme widening of inequality in the 19th century, a massive middle class emerged due to new technological advances, including mass production, in the first half of the 20th century. As a result, the inequality narrowed, leading to the arrival of a mass society. If we believe in market resilience, rather than in the principle of natural selection, the vision of a future human society is not necessarily a dystopian one.

While we may not have to take the dystopian argument in Homo Deus at face value and worry about its consequences, important concerns and uncertainties raised by the argument can be summarized into the following two questions:

(i) How should ordinary people like us, who are exposed to the risk of degenerating into a useless class, lead our lives?
(ii) Does AI (or Homo Deus, who is equipped with increased capabilities with the help of AI) actually pose the risk of wiping out humans?

Of the above two points, (ii) is the issue of how AI will perceive and influence the world, so that point will be discussed later.

(i) is the issue facing people who have lost their existing social roles due to the development of AI and their options for survival for finding their correct place in society. The values which define individuals' relationships with society constitute political philosophy. Political philosophy provides a conceptual framework for discussing and providing criteria on what form social, economic and political systems should take. A philosophy that reflects on what human society should be like also provides indications on how individuals should live within the surrounding social environment.

In other words, question (i) can also be framed as "How can we work out a new political philosophy in the era of AI?" To consider this point, I will review the political philosophy under modern liberalism in following chapters.

December 10, 2021

Article(s) by this author