« Back to Results

Economic Consequences of Artificial Intelligence and Robotics

Paper Session

Saturday, Jan. 6, 2018 2:30 PM - 4:30 PM

Marriott Philadelphia Downtown, Grand Ballroom Salon E
Hosted By: American Economic Association
  • Chair: Erik Brynjolfsson, Massachusetts Institute of Technology

Artificial Intelligence, Automation and Work

Daron Acemoglu
,
Massachusetts Institute of Technology
Pascual Restrepo
,
Boston University

Abstract

We summarize a framework for the study of the implications of automation and
AI on the demand for labor, wages, and employment. Our task-based framework
emphasizes the displacement effect that automation creates as machines and
AI replace labor in tasks that it used to perform. This displacement effect
tends to reduce the demand for labor and wages. But it is counteracted by a
productivity effect, resulting from the cost savings generated by
automation, which increase the demand for labor in non-automated tasks. The
productivity effect is complemented by additional capital accumulation and
the deepening of automation (improvements of existing machinery), both of
which further increase the demand for labor. These countervailing effects
are incomplete. Even when they are strong, automation increases output per
worker more than wages and reduce the share of labor in national income. The
more powerful countervailing force against automation is the creation of new
labor-intensive tasks, which reinstates labor in new activities and tends to
increase the labor share to counterbalance the impact of automation. Our
framework also highlights the constraints and imperfections that slow down
the adjustment of the economy and the labor market to automation and weaken
the resulting productivity gains from this transformation: a mismatch
between the skill requirements of new technologies, and the possibility that
automation is being introduced at an excessive rate, possibly at the expense
of other productivity-enhancing technologies.

What Can Machines Learn, and What Does It Mean for the Occupations and Industries?

Erik Brynjolfsson
,
Massachusetts Institute of Technology
Tom Mitchell
,
Carnegie Mellon University
Daniel Rock
,
Massachusetts Institute of Technology

Abstract

The increased availability of high quality data and rapid advances in machine learning (ML) algorithms have the potential to generate significant economic value in the coming decade. Yet recent estimates suggest that median wage stagnation is in part due to increased automation of routine information processing work and the use of robots has been linked to declines wages and employment for factory workers, raising questions about the distributional effects of more widespread use of ML in the economy. We develop a model in which firms adopt ML and compete with heterogeneous inputs. In this model, some production inputs are complementary to ML while others can be substituted. Worker, firms and industries with complementary investments (e.g. relevant skills, large databases) are well-positioned to grow their value with ML, but the digital nature of ML investments makes these industries susceptible to superstar effects and increased concentration. We develop a taxonomy of task suitable for ML and estimate some implications of our model by applying natural language processing techniques to data from a major online job postings site. We use our taxonomy to predict the adoption of ML applications for skills, jobs, firms, industries, and regions. The potential of ML is widespread, though the current the distribution of ML value is very uneven. Once a task is learned once within a firm, it can be rapidly scaled and applied across the company at near zero marginal cost. Employment at the industry level is affected by changes in the production function as well as changes in market power, with higher value accruing individuals and firms that control essential complements for ML use.

Linking Advances in Artificial Intelligence to Skills, Occupations, and Industries

Rob Seamans
,
New York University
Edward W. Felten
,
Princeton University

Abstract

Prior episodes of automation have led to economic growth and also to many changes in the workplace. In some cases automation has substituted for labor and in other cases automation has complemented labor. We expect that artificial intelligence (AI) will boost economic growth while affecting labor in different ways. The link between AI and labor is complex, however. Our paper provides a method that we believe can help researchers and policy makers to better understand the link between AI and labor. We also demonstrate the method in several applications, including predicting which occupation descriptions will change the most due to advances in AI.

Human Judgment and A.I. Pricing

Joshua Gans
,
University of Toronto
Avi Goldfarb
,
University of Toronto
Ajay Agrawal
,
University of Toronto

Abstract

Recent artificial intelligence advances can be seen as improvements in prediction. We examine how such predictions should be priced. We model two inputs into decisions: a prediction of the state and the payoff or utility from different actions in that state. The payoff is unknown, and can only be learned through experiencing a state. It is possible to learn that there is a dominant action across all states, in which case the prediction has little value. Therefore, if predictions cannot be credibly contracted upfront, the seller cannot extract the full value, and instead charges the same price to all buyers.
Discussant(s)
Benjamin Jones
,
Northwestern University
Shane Greenstein
,
Harvard Business School
Susan Helper
,
Case Western Reserve University
Miguel Villas-Boas
,
University of California-Berkeley
JEL Classifications
  • O4 - Economic Growth and Aggregate Productivity
  • O4 - Economic Growth and Aggregate Productivity