On April 21, 2023 NYU’s Center for the Future of Management hosted a conference focused on the economics of robots at NYU’s Washington DC campus. The in-person conference was attended by approximately 75 participants from academia, industry, non-profit and government organizations. Support for the conference was generously provided by the following organizations: Association for Advancing Automation, Federation of American Scientists, Google, Institute for Progress, Microsoft and NYU’s Stern School of Business. The in-person meeting was preceded by two smaller virtual workshops held in Fall of 2022 that highlighted the “big questions” around the economics of robots. These questions include: 1) potential barriers to robot adoption and use, 2) measurement issues, and 3) the effect of robots on workers and labor.
In his opening remarks Rob Seamans from NYU’s Stern School of Business, Department of Management & Organizations, summarized findings from prior research on the economics of robots. Notably, a number of research papers show that firms adopting robots experience increases in productivity and employment. However, these gains potentially come at the expense of non-adopting firms, raising the question of why these firms don’t also adopt robots.
The first session focused on this puzzle with three presentations that each addressed potential barriers to robot adoption. This session was moderated by Jeff Burnstein from the Association for Advancing Automation. A presentation by Ani Kelkar from McKinsey & Co. shared trends on robot adoption, pointing out that one main reason for adoption is to improve organizational resilience. Nancey Green Leigh from Georgia Tech discussed robot adoption trends in the U.S., highlighting the important role of robot integrators. Stijn Vanormelingen from KU Leuven presented work based on data from Belgium showing that increased robotics adoption leads to an increase in employment but no change in skill mix. Instead, the new skills come from increased use of high skilled indirect labor at integrators.
In the second session, Victor Bennett from the University of Utah moderated a panel on measuring robot quality. The panelists were Sue Helper from Case Western Reserve University, Ron Jarmin from the U.S. Census Bureau, Kim Anderson from Copenhagen Business School and David Byrne from the Federal Reserve. The panel discussed measuring changes in robot quality over time, identifying heterogeneous uses of robots, and other ways of gathering and measuring economic activity.
In the third session, Katya Klinova from the Partnership on AI led a panel covering recent research on generative AI. The panelists were Peter Cihon from GitHub and Pamela Mishkin from OpenAI, who each discussed their recent research papers. The panelists highlighted both the types of tasks most likely to be affected by generative AI and the users who benefit most from these technologies. The panel also discussed the ways in which lessons from robotics research may or may not also pertain to AI.
In the fourth session, Betsey Stevenson from the University of Michigan moderated a set of presentations on the impact of robots on human labor. Lynn Wu from the University of Pennsylvania presented research showing that Canadian firms adopting robots see an increase in employment, but more for production workers than managers. Osea Guintella from the University of Pittsburgh discussed research linking robots to a reduction in worker injuries and an increase in mental health problems. Matthew Beane from UC Santa Barbara discussed nationwide field research on how entry-level warehouse workers built significant skills in jobs that were aggressively deskilled through automation.
The final panel, moderated by Chris Meserole from Brookings, focused on labor policy and workforce development. The panelists were Amanda Ballantyne from AFL-CIO, Claude Dinsmoor from FANUC, Allison Forbes from the Center for Regional Economic Competitiveness, and Anton Korinek from Brookings and the University of Virginia. The panel discussed worker voice and the role of unions, strategies for upskilling workers and building talent pipelines, and reasons why economists and policymakers should be forward-thinking about potential effects of robots and AI on labor.
Conference participants responded to a survey after the event to indicate additional interest in a variety of topics. The following topics were the most popular among survey respondents:
Outcomes. How should we measure productivity (resulting from robot adoption) across industries? Aside from employment and productivity, what other outcomes are important to measure, perhaps patents or some other measure of product innovation?
Barriers to adoption and use. Why do many firms not adopt robots? Is the bottleneck to more adoption due to a lack of hardware or software development? What complementary assets are important for the adoption of robots? For firms that adopt robots, what explains who becomes a “power user” and who becomes a “low user”?
Substitutes vs complements for robots. Which workers are complements or substitutes for robots in the plant-level production process? What are examples where robots have been used to augment workers’ skills? How much of robot adoption is directed at improving product quality instead of reducing labor?
Substitutes vs complements for AI. When does AI complement or substitute for workers?
Policy. What types of policies should be considered to balance increased productivity with workplace disruptions or other tradeoffs from using AI and robots?