‘Robot Overlords’ Job-Stealing Exaggerated: Jackson Hole Paper

Photographer: Justin Sullivan/Getty Images

Autor cites Google Inc. research on driverless cars and onrecognizing images of cats to show the robot takeover is morehype than fact. The Mountain View, California-based company has Close

Autor cites Google Inc. research on driverless cars and onrecognizing images of cats to... Read More

Close
Open
Photographer: Justin Sullivan/Getty Images

Autor cites Google Inc. research on driverless cars and onrecognizing images of cats to show the robot takeover is morehype than fact. The Mountain View, California-based company has

Robots and computers don’t steal as many jobs as some believe, and automation actually benefits many workers, Massachusetts Institute of Technology Professor David Autor will say in a presentation at a Federal Reserve conference.

A key reason humans aren’t obsolete yet is that simple tasks such as visually identifying a chair, which any child can do, aren’t so easy for engineers to teach to computers, Autor said in a paper prepared for presentation today at the Kansas City Fed’s annual symposium in Jackson Hole, Wyoming.

“Journalists and expert commentators overstate the extent of machine substitution for human labor and ignore the strong complementarities that increase productivity, raise earnings, and augment demand for skilled labor,” he wrote. “Challenges to substituting machines for workers in tasks requiring flexibility, judgment, and common sense remain immense.”

Autor, 50, is a top U.S. labor market researcher who focuses on shifts in the labor market such as job polarization, the concentration of job growth at the high and low ends of the labor market. His paper is one of four being discussed at the conference.

Autor cites Google Inc. (GOOG) research on driverless cars and on recognizing images of cats to show the robot takeover is more hype than fact. The Mountain View, California-based company has conducted more than 300,000 miles of driverless car testing with vehicles. They navigate by collecting real-time sensor data and comparing it to pre-loaded maps that specify exact locations for roads and signs, while adapting to obstacles such as people and cars.

Google Cars

Without the maps, they can’t drive in terrain they don’t recognize. “While the Google car appears outwardly to be as adaptive and flexible as a human driver, it is in reality more akin to a train running on invisible tracks,” Autor writes.

Google researchers also taught computers to identify images of cats on YouTube. It took a neural network of 16,000 processors. “Even exemplary accomplishments in this domain can appear somewhat underwhelming,” Autor writes.

His paper also observed that, while International Business Machines Corp.’s Watson supercomputer beat “Jeopardy!” quiz show champions, it wasn’t perfect. In the category of U.S. Cities, the question was: “Its largest airport was named for a World War II hero; its second largest, for a World War II battle.” Watson said Toronto. (The correct response is Chicago.)

Teenage Programmer

Autor learned to program computers as a teenager and earned his undergraduate degree in psychology, with a concentration in computer science, at Tufts University in Medford, Massachusetts. He got a master’s in public policy at Harvard University’s John F. Kennedy School of Government, writing a thesis titled “Charting the Impacts of the Computer Revolution: New Evidence from the Current Population Survey.”

He went on to earn a doctorate in public policy at the Kennedy School, in Cambridge, Massachusetts, with a thesis titled “Essays on the Changing Labor Market: Computerization, Inequality, and the Development of the Contingent Work Force.”

In 1999 he joined the faculty at MIT, also in Cambridge, and now is associate head of the economics department.

Autor, attending the Kansas City Fed’s symposium for the first time, wrote the paper at the request of the reserve bank hosting the three-day gathering of central bankers and economists, titled “Re-evaluating Labor Market Dynamics.”

He writes in his paper to be presented today that “popular and academic discussion of the potentially dire consequences of automation for employment has accelerated,” citing 2013 stories by journalist Noah Smith in The Atlantic warning of “the end of labor” and Kevin Drum in Mother Jones predicting “robot overlords” will steal jobs from workers.

Artificial Intelligence

He also cites his MIT colleagues Erik Brynjolfsson and Andrew McAfee. Their 2011 book “Race Against The Machine” catalogs how artificial intelligence gains allow computers to perform tasks only humans could before.

“Where are these robot overlords?” Autor wrote. “And if they are not here already -- and all outward appearances suggest that they are not -- should we expect their imminent arrival?”

While computerization has progressed into areas considered off limits a few years ago earlier, such as driving, reading legal documents, and doing farm labor, automated systems still lack flexibility, Autor wrote. Auto factories use robots to install windshields, yet only humans install replacements “because removing a broken windshield, preparing the windshield frame to accept a replacement, and fitting a replacement into that frame demand far more real-time adaptability than any contemporary robot can approach.”

Complementing Humans

Autor said there’s a “long history of leading thinkers overestimating the potential of new technologies to substitute for human labor and underestimating their potential to complement it.”

Focusing on what’s lost in mechanization “misses the central economic mechanism,” in which productivity growth raises the value of work only humans can do because machines both substitute for and complement human labor. Mechanization plays a positive role in some industries, such as construction, according to the paper.

“By historical standards, contemporary construction workers are akin to cyborgs,” Autor wrote. “Augmented by cranes, excavators, arc welders, and pneumatic nail guns, the quantity of physical work that a skilled construction worker can accomplish in an eight-hour workday is staggering.”

As automation substitutes for human construction labor, fewer workers are needed to accomplish a task than 50 years ago, Autor says, yet “construction workers have not been devalued by this substitution.”

Employment Gains

Even though many jobs have become obsolete, such as auto factories sealing the fate of blacksmiths, stable hands, and other horse-related trades, Autor says the losses were later “more than offset” by employment gains. He cited the share of the U.S. work force employed in agriculture, which fell from 41 percent in 1900 to 2 percent in 2000.

“It is unlikely, however, that farmers at the turn of the 20th century could foresee that 100 years later, health care, finance, information technology, consumer electronics, hospitality, leisure and entertainment would employ far more workers than agriculture,” Autor says. “Arguably, we stand at a similar moment today.”

To contact the reporter on this story: Jeff Kearns in Washington at jkearns3@bloomberg.net

To contact the editors responsible for this story: Chris Wellisz at cwellisz@bloomberg.net Mark Rohner, Brendan Murray

Press spacebar to pause and continue. Press esc to stop.

Bloomberg reserves the right to remove comments but is under no obligation to do so, or to explain individual moderation decisions.

Please enable JavaScript to view the comments powered by Disqus.