TalaMindLLC

Following is the discussion of "technological unemployment" given in Dr. Jackson's book "Toward Human-Level Artificial Intelligence", published by Dover in 2019. See that book for the references to other authors given below. Almost all of the following somewhat lengthy discussion is also given in the third edition of his book "Introduction to Artificial Intelligence", also published by Dover in 2019:

In 1930, Keynes defined 'technological unemployment' as unemployment caused by technology eliminating jobs faster than it creates new jobs. He warned it would be a significant problem for future generations.

In 1983, Leontief, Duchin, and Nilsson each wrote papers about the potential for automation and AI to cause long-lasting unemployment. Leontief (1983a,b) reasoned the use of computers to replace human mental functions in producing goods and services would increasingly reduce the need for human labor. Nilsson (1983) predicted AI would significantly reduce the total need for labor, particularly for white-collar and service sector jobs. Duchin (1983) discussed methods for widely distributing incomes without paychecks.

In the past two decades, several authors have warned about this potential problem and suggested possible solutions. They include Albus (2011), Brain (2013), Brynjolfsson and McAfee (2011), Ford (2009), Reich (2009 et seq.), Rifkin (1995 et seq.), and others. So, several economists (Brynjolfsson, Duchin, Leontief, McAfee, Reich, Rifkin) and computer technologists (Albus, Brain, Ford, Nilsson) have discussed this problem and developed similar viewpoints. To be concise in referring to these authors, they will here be called Leontief-Duchin-Nilsson (LDN) theorists, focusing only on their arguments regarding technological unemployment, automation, and AI—they may disagree about other topics. It would be incorrect to call them Keynesian economists, since this term refers to Keynes' theories more broadly. Nor is it accurate to call them Luddites or neo-Luddites, because they do not advocate halting technological progress.

Economists in general disagree about whether technological unemployment can have widespread and long-lasting effects on workers and the economy. Many economists have considered it is not a significant problem, arguing that workers displaced by technology will eventually find jobs elsewhere, and the long-term effect on an economy will be positive. However, Leontief, who was awarded the Nobel Memorial Prize in Economic Sciences in 1973, wrote that it is not valid to assume that someone who loses a job due to technological progress will always be able to find another job, even after retraining. Brynjolfsson and McAfee (2011) wrote that no economic law says technological progress will automatically benefit most of the people. A large majority of the people in a nation can have reduced wealth as a result of technological progress, even if overall wealth increases.

Since it is beyond the scope of this book to resolve disputes about theories of economics, the views of LDN theorists can at most be presented tentatively. Those writing in the past two decades appear to roughly agree at least implicitly, on the following points for the problem of technological unemployment:

1. In the next several decades of the 21st century, automation and AI could lead to technological unemployment affecting millions of jobs at all income levels, in diverse occupations, and in both developed and developing nations. This could happen with current and near-term technologies, i.e. without human-level AI. It has already occurred for manufacturing, agriculture, and many service sector jobs.

2. It will not be feasible for the world economy to create new jobs for the millions of displaced workers, offering equivalent incomes producing new products and services.

3. Widespread technological unemployment could negatively impact the worldwide economy, because the market depends on mass consumption, which is funded by income from mass employment. LDN theorists vary in discussing and describing the degree of impact.

4. The problem is solvable by developing ways for governments and the economy to provide alternative incomes to people who are technologically unemployed. LDN theorists have proposed methods for funding and distributing alternative incomes.

5. The problem can and should be solved while preserving freedom of enterprise and a free market economy.

6. The problem cannot be solved by halting or rolling back technological progress, because the world's population depends on technology for economic survival and prosperity.

7. Solutions to the problem could lead to greater prosperity, worldwide. LDN theorists vary in describing potential benefits: Nilsson (1984) envisioned automation and AI could provide the productive capacity to enable a transition from poverty to a "prosperous world society". Ford (2009) suggested extension of alternative incomes to people in poverty could create market demand causing a 'virtuous cycle' of global economic growth.

Based on the arguments of LDN theorists, the possibility that AI could help eliminate global poverty may be considered a 'potential best case event' for the economic risks and benefits of AI.

To summarize: If technological unemployment is a major economic problem, then global prosperity could require developing an economic system that provides alternative incomes to people who are technologically unemployed. The challenge could be to develop ways of funding a universal basic income while preserving freedom of enterprise and economic stability, and controlling monetary inflation.

Wray (1998) discussed how the federal government could be the 'employer of last resort' in the US economy, using 'modern monetary theory' (MMT). Work on repairing and upgrading the nation's infrastructure could offset or postpone technological unemployment for perhaps millions of workers for the next few decades. Much work could be funded in the service sector, providing care for the environment, people, and communities. MMT funding could also provide income for extended periods of time to people who are technologically unemployed. Such funding would minimize taxation and avoid over-taxation for all income levels, and enable capitalism and free markets to continue supporting economic growth. It could reduce the need for state and local taxes. (N.b: MMT and LDN theory originated separately.)

Technological Unemployment