COMPUTING AND AI: HUMANISTIC PERSPECTIVES FROM MIT
Economics | Nancy L. Rose and David Autor
Nancy Rose, Kindleberger Professor of Applied Economics; Head, MIT Dept of Economics; photo by Allegra Boverman
"The intellectual affinity between economics and computer science traces back almost a century, to the founding of game theory in 1928. Today, the practical synergies between economics and computer science are flourishing. We outline some of the many opportunities for the two disciplines to engage more deeply through the new MIT Schwarzman College of Computing."
— Nancy L. Rose and David Autor, MIT Professors of Economics
Nancy L. Rose is the Charles P. Kindleberger Professor of Applied Economics and head of the MIT Department of Economics, where her research and teaching focus on industrial organization, competition policy, and the economics of regulation.
David Autor is the Ford Professor of Economics and co-director of the MIT Task Force on the Work of the Future. His scholarship explores the labor market impacts of technological change and globalization, earnings inequality, and disability insurance and labor supply.
. . .
Q: What are examples of domain knowledge, perspectives, and methods from economics that should be integrated into the research and curriculum of the MIT Schwarzman College of Computing (SCC) and why?
The intellectual affinity between economics and computer science traces back almost a century, to the founding of game theory in 1928. Today, the practical synergies between economics and computer science are flourishing. The following are some of the many opportunities for the two disciplines to engage more deeply through the SCC.
Preparing for the work of the future
Through the MIT Work of the Future Task Force, launched by MIT President L. Rafael Reif in 2018, MIT has forged a leadership role in responding to the potentially disruptive labor market consequences of rapid advances in artificial intelligence (AI) and robotics. The task force will map the technological and policy landscape confronting the United States and other countries, and consider how institutions, governments, and individuals can embrace the opportunities and surmount the challenges posed by disruptive changes in employment and the needs of the workplace.
The field of market design, pioneered by Nobel Laureate in economics Alvin Roth, pairs tools from traditional game theory with computer science to engineer market structures that yield socially desirable outcomes — often in settings where participants have complex preferences and prices may be specifically disallowed. Advances by MIT Economics faculty member Parag Pathak have improved the assignment of public school slots to students. His colleague Nikhal Agarwal uses these tools to analyze the allocation of kidneys to patients needing transplants.
Several MIT classes already address core ideas from economic theory and apply methods pioneered by economists, including classes in game theory, algorithmic game theory, and market design. We anticipate that many more opportunities for cross-disciplinary collaboration in both teaching and research will arise organically within SCC.
In a related vein, e-commerce has opened new markets and fundamentally transformed existing ones. Research is needed to understand the economic potential of these markets and to contribute to their design. Economics faculty members Glenn Ellison and Sara Fischer Ellison are leaders in theoretical and empirical research on online markets and their interaction with brick and mortar enterprises. Numerous computer scientists are active in these areas as well, and papers in computer science and economics routinely cite one another.
David Autor, Ford Professor of Economics at MIT; photo by Peter Tenzer
"Invest in research on market design and industrial organization to ensure that the benefits from innovations in online markets accrue to consumers as well as firm productivity and profits."
A critical issue in market design is assuring robust competition. Even as e-commerce has ushered new markets into existence, it has seemingly catalyzed market dominance by a handful of frontier technology firms. Economics offers sophisticated, time-tested models that illuminate the sources of market power and quantify the economic distortions that flow from it. When firms use personalized consumer data to target and price their services, this can either create value for consumers or capture value that consumers would have otherwise received. MIT economics faculty are striving to understand how rapid changes in market structure arising from information technology are reshaping productivity, profits, and consumer welfare; how technology choices influence this; and what the implications are for public policy.
Building fairness, security, and privacy into algorithmic design
The explosion of commercial AI has made pressing the question of algorithmic fairness and, more ominously, the potential for algorithmic bias, where machines adopt the prejudices and cognitive defects of their human role models. Issues of differential privacy and data imputation are also vital to empirical research in economics and areas of study in econometric methods.
Obtaining causal inference from big data
The algorithmic tools and computing capacity for processing vast quantities of unstructured data have evolved tremendously in recent decades, but the tools for drawing causal inference from those data remain in their infancy. Most commercial “big data” applications focus on the comparatively simple task of prediction: What product will a consumer likely buy; what link will they click next, etc. Prediction is the least demanding statistical problem because it does not require knowledge of cause and effect, only of association. MIT’s Victor Chernozhukov advances the frontiers of both prediction and causal interference methods.
Recognizing the centrality of causal inference to scientific understanding, economists have devoted substantial mindshare to developing and field-testing tools for credibly answering cause-and-effect questions, a mission elevated in the economics profession by MIT’s Josh Angrist among others.
This toolkit includes, of course, the classical technique of randomized controlled experimentation.Unique to economics, it also encompasses a set of statistical methods for obtaining causal inference from observational data, including quasi-experiments, matching estimators, regression discontinuity techniques, synthetic control methods, and instrumental variables models. MIT’s econometricians and applied economics faculty are on the leading edge of developing and applying these tools.
Modeling interactions on networks
While network congestion has traditionally been viewed as an engineering problem by computer scientists, collaborative research between computer sciences and economists has demonstrated the degree to which congestion is a classic "tragedy of the commons" problem: Absent positive usage prices, network users have every incentive to consume bandwidth without regard to the externalities imposed on others and to increase their network demands as congestion rises. The study of market interactions on networks is another area where faculty members in economics and computer science are collaborating, among them Daron Acemoglu and Asu Ozdaglar.
Algorithmic fairness; collage by SHASS Communications
"Support research on fairness, expertise, cognitive biases, and preference in machine-supported and machine-delegated decision-making: The explosion of commercial AI has made pressing the question of algorithmic bias, where machines adopt the prejudices and cognitive defects of their human role models."
Q: What exciting, meaningful opportunities do you see for advancing the field of economics through the SCC?
We hope that as the SCC moves rapidly from a pathbreaking idea to a dynamic institution connecting and augmenting the five schools of MIT, the new college will seed research and educational investments that join computer science and economics across multiple domains:
• Supporting research on fairness, expertise, cognitive biases, and preference in machine-supported and machine-delegated decision-making.
• Investing in research on market design and industrial organization to ensure that the benefits from innovations in online markets accrue to consumers as well as firm productivity and profits.
• Pursuing scholarship at the intersection of data science, econometrics, and causal inference.
• Building depth in network science, algorithmic game theory and mechanism design, and online learning.
• Institutionalizing MIT’s research capacity, educational efforts, and policy expertise in studying and shaping the future of work.
• Developing tools for rapid, cost-effective, and ongoing education and retraining for current and future cohorts of workers.
Through these investments, the SCC can advance its mission of accelerating research and innovation in computing, building expertise in computing’s societal impacts and ethical implications, and educating leaders to shape this rapidly unfolding future.
Nancy Rose appointed head of MIT Department of Economics
Distinguished scholar and MacVicar Faculty Fellow is also an accomplished public servant.
MIT economist Nancy Rose to take Department of Justice position
Expert in regulation and market competition will take leave to spearhead DOJ’s economic analysis.
Why cities aren’t working for the working class
David Autor’s latest research shows how economic polarization stems from urban job loss.
David Autor on global trade and political polarization
Study finds relationship between U.S. job losses due to trade, and political polarization in Congress.
David Autor named 2018 MacVicar Fellow
The Institute's highest undergraduate teaching honor
MIT report examines how to make technology work for society
Task Force on the Work of the Future calls for bold public and private action to harness technology for shared prosperity.
MIT launches Task Force on the Work of the Future
Institute-wide effort will study the evolution of jobs in an age of technological advancement.
Two sciences tie the knot
A new major combining computer science and economics will prepare students for designing the virtual marketplaces of the future.
At MIT forum, economists evaluate the consequences of increasing inequality in America, and suggest solutions.
Series prepared by MIT SHASS Communications
Office of Dean Melissa Nobles
MIT School of Humanities, Arts, and Social Sciences
Series Editor and Designer: Emily Hiestand, Director of Communications
Series Co-Editor: Kathryn O'Neill, Associate News Manager
Published 23 September 2019