Exploring the Ethics of Automation


In May, MIT economist Daron Acemoglu spoke during a webinar organized at MIT Sloan titled “Ethical Automation: Shaping the Future of Work.”

But despite the event’s title, Acemoglu cautioned that “ethical automation is not what we have been doing” in the U.S. economy in recent years. Instead, he argued, business has been pursuing a path of “excessive automation.”

Acemoglu, who is an Institute Professor in the MIT Department of Economics and wrote the best-selling book “Why Nations Fail” with James Robinson, spoke as part of a webinar cosponsored by the MIT Institute for Work and Employment Research’s Good Companies, Good Jobs Initiative; the MIT Sloan People and Organizations Club; and the MIT Sloan Office of Student Life. The webinar, which also featured Amanda Hauser DeHaven, Regional Director of Strategic Projects at Kaiser Permanente, was the fifth installment of the groups’ “Redefining Management: Leadership for Social Progress in Troubled Times” 2020-2021 virtual speaker series. The event was moderated by Nick Brenner, co-president of the MIT Sloan People & Organizations Club. Barbara Dyer, a Senior Lecturer at MIT IWER and the Executive Director of the Good Companies, Good Jobs Initiative at MIT Sloan, introduced the event.

Acemoglu began his remarks by explaining that “not all is well in the U.S. labor market,” with “huge inequality” in weekly earnings by education level emerging from the late 1970s on. In recent decades, real “median wages, especially for men, are stagnant,” Acemoglu observed, and workers with less than a four-year college education have been experiencing a decline in real wages. While other advanced economics have also been experiencing increases in inequality and declines in middle-class jobs, this pattern of declining real wages for the least-educated workers is almost unique to the U.S., he said, and has major economic and social costs for the country.

There are many causes for these trends, according to Acemoglu, “but technological changes have played a major role.” In his work with Pasqual Restrepo of Boston University, Acemoglu has identified two types of technologies: some, such as automation, simply displace people and others create new tasks for workers that replace old ones.

In the U.S. in recent years, Acemoglu argued, there has been a strong bias toward automation, leading to what he calls “excessive automation.” He said this bias toward automation in the U.S. has several causes: business’s focus on cost-cutting in the face of global competition, the business models of large technology companies, and a U.S. tax code that “was always very anti-labor” and has become much more so in recent years. Software and equipment, according to Acemoglu, now face an effective tax rate in the neighborhood of 5% in the U.S. while labor is taxed at more than 25%. “There’s a huge hidden incentive in the tax system” for firms to automate, even when it’s not productive, Acemoglu explained.  As a result, firms are overautomating, in Acemoglu’s opinion, and society is not really getting many productivity benefits from all this automation.

However, this path is not inevitable, according to Acemoglu. “The direction of technology is a choice,” he said, and technology that helps workers rather than replacing them tends to result in good jobs and high wages. He called for new institutional frameworks that reexamine what we’re doing with automation, and regulation of technology is something that needs to be rethought.  Otherwise, he predicted, we are headed toward a two-tiered society.

Daron Acemogul | Institute Professor
“Ethical automation would be automation that is limited, takes its effects on a broad set of stakeholders into account and is counterbalanced by other opportunities for good jobs,”

After Acemoglu spoke, Hauser DeHaven described her organization’s approach to involving workers in the design and implementation of a set of new technology-enabled clinics in southern California—a project that was the subject of a 2020 working paper published by the MIT Task Force on the Work of the Future. Kaiser Permanente is a California-based health care system that has one of the longest-enduring and largest-scale labor-management partnerships in the United States.

Hauser DeHaven described her organization’s approach as not one of implementing technology for technology’s sake but instead implementing technology that will enhance communication on Kaiser Permanente’s care teams and between the care teams and patients. She said that part of the process for planning the new clinics was interviewing employees about their technology preferences and incorporating their feedback and concerns.  It was important, Hauser DeHaven said, to give employees time to practice using the new technologies. Kaiser Permanente also had to renegotiate some job descriptions with its labor partners, she said.