A shocking piece of news struck the world of “new and old” in the first week of 2017—AlphaGo, an artificial intelligence (AI) program developed by Google DeepMind in London to play the board game Go, had secretly won by a landslide a series of online Go matches against world class human masters of the game. While the developers of AlphaGo celebrated this remarkable achievement, many experts were deeply disturbed by the potential implications of AI for human existence. Renowned entrepreneur, engineer, and inventor Elon Musk, despite his own achievements at the frontier of technology: solar power, electric vehicles, and space rocketry, ominously described the development of AI as a rival to human intelligence as “summoning the demon” and potentially the biggest threat facing humanity. (The Economist) Martin Rees, founder of the Centre for the Study of Existential Risk at the University of Cambridge, shared similar misgivings.
For all the benefits technology has brought to our daily lives in the form of telecommunications, transportation, healthcare, and so on, it might be hard to imagine how the fruits of human innovation could backfire and hurt us. The prospect of humans losing control to AI seems so far-fetched as to be the stuff of computer-generated imaginings in science fiction movies such as the Terminator film franchise and I, Robot.
While a recent commentary published by Buddhistdoor Global* discusses the ethical implications of AI, this article focuses on the economic perspective. Early economists such as Adam Smith applauded the division of labor as an important market development that would allow for specialization and exchange. Nonetheless, some contemporary political economists have criticized division of labor for treating humans as machines and eventually replacing obsolete human labor with actual machines. Karl Marx is well known for his strong criticism of capitalism. He argued that increasing specialization only degrades the worker’s skills and spirit through a process in which workers are increasing alienated from their work and the end product. Ernst Schumacher, a pioneer in Buddhist economics, also criticized the ruthless division of labor as making work meaningless and boring, and indicative of “a greater concern with goods than with people, an evil lack of compassion and a soul-destroying degree of attachment to the most primitive side of his worldly existence.” (Schumacher 1984, 45)
The rise of the machines and the potential threat to human existence has been part of our history starting with the use of primitive stone implements, to fashioning metallic tools, to the invention of the steam engine, electricity, mobile communications, computers, and so on. It has been a process of first dividing human capacity to act and think in definable, repetitious, and replicable steps, then introducing machine automation to mimic the human process, and finally, replacing workers with machinery or computers. A recent report by management consulting firm McKinsey & Company explores “Where machines could replace humans—and where they can’t (yet).” Its analysis suggests that automation will have a substantial impact on “almost all jobs to a greater or lesser degree.” In some cases, automation may completely eliminate some job functions, particularly in manufacturing industries, although even knowledge-based industries are at risk.
The McKinsey report also argues that technical feasibility is an important consideration in accessing the potential of automation. Technical feasibility, in turn, could be evaluated by work activities instead of occupations, i.e. by understanding job functions instead of industries. For example, predictable physical work has a much higher technical feasibility for automation (78 per cent) than unpredictable physical work (25 per cent). The most difficult job functions to automate are those involving management and human resources development (9 per cent), and those applying expertise in decision-making, planning, or creative work (18 per cent). In essence, job functions that involve deep expertise, activities that are difficult to clearly define, and complex human interaction have a much lower levels of technical feasibility for automation.
Although AlphaGo’s victory is a showcase for the seemingly limitless potential of AI—this software was able to replicate and supersede certain aspects of intuitive thinking of a grand master in the game of Go—we should never underestimate uniquely human capabilities and potential. Taking computer-vision systems as an example, scientists such as Li Fei-Fei at Stanford University have harnessed years of research, machine learning, and big data to train a super computer to visually distinguish specific parts of an object in a complex setting—even though this is as simple as identifying various items on a coffee table. From a Buddhist perspective, human ability in mental proliferation and conceptualization is indeed powerful. The ability to correctly differentiate context within a pattern is still beyond the reach of AI. Incredible as it may seem, AI still has difficulty distinguishing “a bronze statue of a warrior riding on a horse in front of a building” from “a man riding a horse down a street next to a building.” (TED) It also lacks the ability to appreciate the heartfelt story expressed by an apparently emotionless two-dimensional family photo.
Machines are not taking any rests in the race to catch up with the human ability to conceptualize and proliferate ideas. Still, human mental capacity can continue to lead this race by transcending the bondage and defilement of greed, hatred, and ignorance. Instead of focusing our efforts and energies on an endless game of consumer satisfaction, we could deploy the additional wealth and time accumulated through the application of technology to further develop our own mental capacities. We could devote more resources to our own development as human beings. What if we spend more time and resources on education and healthcare, which would contribute to our physical and mental health? What if we invest in areas in which we have a comparative advantage—human interaction and the deeper levels of consciousness that enable mindfulness, morality, and wisdom? Certainly, it may take a while for AI to develop full visual recognition. It will take even longer for machines, if it is possible at all, to appreciate concepts such as justice, emptiness, and compassion. We can maintain our leadership in this race and keep our jobs, or allow our humanity and thousands of years of wisdom and learning to waste and wither away.
Schumacher, Ernst F. 1984. Small is Beautiful: Economics as if People Mattered. London: Sphere Books. Original edition 1973.
Where machines could replace humans—and where they can’t (yet) (McKinsey Quarterly)
Rise of the machines (The Economist)
How we’re teaching computers to understand pictures (TED)
What the AI behind AlphaGo can teach us about being human (WIRED)
Google secretly squared off its AI against leading Go players and it won by a landslide (TNW)
Karl Marx 1844: Economic and Philosophic Manuscripts of 1844 (Marxists Internet Archive)