AI and the Future of Work

Imagine an observer from another galaxy descends to Earth and sees entities called humans and entities called machines interacting with each other. 

How would the observer answer the question “Which entity controls the other?”

In a recent paper, University of Virginia Darden School of Business Professor Anton Korinek presents the thought experiment as a part of a broader consideration meant to put society’s increasingly codependent relationship with technology into context.

“Determining which entity is in control would not be simple,” said Korinek, who joined Darden at the start of the 2018–19 academic year and holds a joint appointment with the UVA Department of Economics.

“Sometimes the machines start beeping, and we tend to them and follow their instructions and commands. When they cry, we plug them into little outlets and give them food again,” Korinek said. “The observer from another planet would probably think that both humans and machines are moderately intelligent entities with different strengths and live in some sort of mutual symbiosis, as biologists would call it.”

According to Korinek, people tend to take a distinctly human-centric view of the agents of activity in our economy ― putting humans at the center ― while discounting the rising impact of the machines.

Near-Term Change at Work

Artificial intelligence (AI) is something of a catchall term for the ability of machines to perform tasks that would traditionally require human intelligence. Conditioned by decades of science fiction movies and novels, people may think of AI as a technology of the future, but in reality millions are already using applications powered by AI each day, from personal digital assistants like Apple’s Siri and Amazon’s Alexa to recommendation engines used by companies such as Spotify and Netflix. 

A 2018 report from the McKinsey Global Institute examines five categories of AI: computer vision, natural language, virtual assistants, robot process automation and advanced machine learning. The report says the rise of AI will likely usher in “arguably unprecedented” changes in the way people live and work.

And according to the World Economic Forum’s 2018 Future of Jobs Report, AI is one of the four specific technological advances driving change in business in the period from 2018 to 2022, joining widespread high-speed mobile internet, big data analytics and cloud technology as significant disruptors to the status quo.

To give a sense of the near-term disruption: The World Economic Forum report predicts 42 percent of existing work task hours will be performed by machines by 2022, up from 29 percent in 2018. And as a sign of the complexities of the issue, the report lists “Advances in Artificial Intelligence” on both the lists of trends that will positively and negatively impact business growth up to 2022.

“There are complex feedback loops between new technology, jobs and skills,” according to the World Economic Forum report. “New technologies can drive business growth, job creation and demand for specialist skills, but they can also displace entire roles when certain tasks become obsolete or automated.”

Long-Term Change in Life

Korinek, who has turned his focus to the likely economic fallout from an AI-powered age of automation, is asking what it means for the future of the economy, work and humanity. He says he is not optimistic about the long-term outlook for any of the three, barring some sort of course correction.

“I think there is a serious risk that, as new technology and automation develop further and further, we will actually end up in a situation in which at first a small, then growing, fraction of the population won’t find gainful employment anymore,” said Korinek, who will teach the elective “Artificial Intelligence and the Future of Work” this academic year. Yesterday’s assembly line worker replaced by a machine is tomorrow’s The Washington Post journalist or Uber driver displaced by writing and driving AI. 

In his 2017 paper “Artificial Intelligence and Its Implications for Income Distribution and Unemployment,” co-authored with 2001 Nobel laureate Joseph Stiglitz, Korinek lays out tactics to assist those who may lose out to AI-spurred automation, including a model of taxation that could be used for compensation, changes to patent law — shortening the term of patent protection could more quickly benefit society through lower prices, once the innovator has already profited — and interventions the government could take to assist workers and insure them against the adverse effects of innovation. Such interventions might mean retraining the labor force, taking advantage of public policy’s influence on the growing service sector of the economy (such as education and health care) or reserving resources valuable to both people and AI — perhaps extending humans permanent usage rights to land and energy.

A society unwilling to grapple seriously with questions of increased automation may see large swaths of the populace resist attempts at innovation efforts, the authors wrote.

What Defines Us as Humans

Although increased efficiencies and automation in means of production have been a constant in the development of the modern working world since the Industrial Revolution, Korinek said the coming epoch feels distinctly different from previous eras.

“Up until now, machines have replaced our physical strength, but they have allowed us to leverage our intellect and intelligence,” Korinek said. “AI is really coming for what has always defined us as humans. It’s coming for our last bastion of advantage over the machines. And that changes the game fundamentally.”

It’s an issue that has been at the forefront of much of Professor Ed Hess’ work, whose 2017 book, Humility Is the New Smart: Rethinking Human Excellence in the Smart Machine Age, says that humans will need to excel at doing the types of thinking and the emotional skills that AI won’t be able to do well.

Korinek’s latest project considers what it means for an economy in which humans and machines, or “artificially intelligent agents,” are increasingly interacting on an equal footing. Korinek’s work shows that while the human-driven economy has been relatively stagnant for the last decade, the machine-driven economy has been booming, showing double-digit growth for the past two decades. It’s a disparity that does not show signs of abating.

“At first I thought it was a completely futuristic project that was important because we are moving so fast in that direction. But the more I’m working on it, the more I notice that we are already seeing in the present several of the implications that my model of artificially intelligent agents would predict for the future,” Korinek said. “That’s a little bit scary.”

Anton Korinek co-authored “Artificial Intelligence and Its Implications for Income Distribution and Unemployment,” which appeared in The Economics of Artificial Intelligence: An Agenda, with Joseph E. Stiglitz, Nobel laureate in economics and University Professor at Columbia University.