IDSS Distinguished Seminars

Title: Designing AI for Racial Equity: Translating Ethics into Practice

Abstract: Among the principles for ethical and responsible AI none is more prominent than the “fairness and non-discrimination” principle, that is, the idea that data-based systems like AI/ML should avoid unjust impacts on people, especially those from historically marginalized populations. But even as this principle is gaining greater academic, industry, and public attention a core challenge persists: how do we operationalize these ideas in the design, development, and deployment of AI/ML? More specifically, how do we build models, systems, and practices that realize the goals of this principle in tangible ways?

In this talk, I address this question through a racial equity framework. That is, a framework that emphasizes the need to design and deploy these systems in ways that intentionally seek to eliminate the impacts of structural racism. The talk frames this call to action as both a technical and social challenge. As a result, it requires dynamic approaches and forms of expertise. From a technical perspective this involves, among other things, thinking strategically about datasets, data governance, and modeling. From a social perspective this involves, for example, developing a domain specific understanding of structural racism and engaging diverse stakeholders. In the talk, I will also discuss how the Good Systems grand challenge at the University of Texas is partnering with the city of Austin and other entities to better understand how to deploy data-based systems to advance racial equity in key life domains such as policing and health.

About the speaker: S. Craig Watkins is the Ernest A. Sharpe Centennial Professor at the University of Texas at Austin and the Founding Director of the Institute for Media Innovation​. His research focuses on the impacts of media and data-based systems on human behavior, with a specific concentration on issues related to systemic racism. He is the author of six books and several articles and book chapters examining the intersections between race, technology, and society. His research also considers how diverse communities seek to adopt and deploy technology in innovative ways that address data literacy, civic life, and health. This work has been supported by the MacArthur Foundation.

Currently, Watkins is leading a team that will address the issue of artificial intelligence and systemic racism in a new six-year program funded by the Office of Vice President for Research at the University of Texas at Austin. The team will focus on the broad and fundamental scientific challenge of achieving racially equitable AI, while being grounded in testing the applicability of specific methods, models, processes, and procedures in critical domains like health and transportation. A key component of the research is to examine how various stakeholders—developers of technologies, the private and public sectors, and citizens—can work to create a more equitable AI future.

Register now: We are required to collect contact information for this event, including name, cell phone number and email address. This information is for contact tracing purposes only. Please provide your contact information here if you plan on attending the seminar.

Title: Designing Equitable Algorithms for Criminal Justice and Beyond

Abstract: Machine learning methods are increasingly used to model risk in criminal justice, banking, healthcare, and other high-stakes domains. These new tools promise gains in accuracy, but also raise challenging statistical, legal, and ethical questions. In this talk, I’ll describe the dominant axiomatic approach to fairness in machine learning, and argue that common mathematical definitions of fairness can, perversely, lead to discriminatory outcomes in practice. I’ll then present an alternative, consequentialist perspective for designing equitable algorithms that foregrounds the inherent tension between competing concerns in many real-world problems.

About the speaker: Sharad Goel is a Professor of Public Policy at Harvard Kennedy School. He looks at public policy through the lens of computer science, bringing a computational perspective to a diverse range of contemporary social and political issues, including criminal justice reform, democratic governance, and the equitable design of algorithms. Prior to joining Harvard, Sharad was on the faculty at Stanford University, with appointments in management science & engineering, computer science, sociology, and the law school. He holds a bachelor’s degree in mathematics from the University of Chicago, as well as a master’s in computer science and a doctorate in applied mathematics from Cornell University.

Abstract: Migration is one of the defining challenges of the 21st century. Many countries have experienced a stark increase in the size and diversity of their immigrant and refugee populations. Reeling and reactive, governments are struggling to facilitate their successful integration. Policymakers responding to the integration challenge often don’t know where to start or what programs work best, and there’s little solid evidence and expertise to guide them. In this talk, I will describe some of the efforts that my research group at the Stanford Immigration Policy Lab (IPL) has undertaken to make headway against these obstacles. This involves our collaborative efforts to build an evidence base for what works and what doesn’t and our exploration into leveraging data science and behavioral insights to create algorithms and digital tools to facilitate better decision making.

About the speaker: Jens Hainmueller is a Professor of Political Science and, by courtesy, a Professor of Political Economy at the Graduate School of Business. He is the co-director of the Stanford Immigration Policy Lab. His research lab works with governments and NGOs around the world to leverage data science, evidence, and innovation to improve policies involving immigrants and refugees.

Zoom meeting ID: TBD
Join Zoom meeting: TBD
YouTube livestream: TBD

Abstract:

We examine a determinant of cultural persistence that has emerged from a class of models in evolutionary anthropology: the similarity of the environment across generations. Within these models, when the environment is more similar across generations, the traits that have evolved up to the previous generation are more likely to be optimal for the current generation. In equilibrium, a greater value is placed on tradition and there is greater cultural persistence. We test this hypothesis by measuring the variability of different climatic measures across 20-year generations from 500–1900ce. Employing a variety of tests that use a range of samples and empirical strategies, we find that populations with ancestors who lived in environments with more cross-generational instability place less importance on maintaining tradition today and exhibit less cultural persistence.

About the speaker:

Nathan Nunn is Frederic E. Abbe Professor of Economics at Harvard University. Professor Nunn’s primary research interests are in political economy, economic history, economic development, cultural economics, and international trade. He is an NBER Faculty Research Fellow, a Research Fellow at BREAD, a Faculty Associate at Harvard’s Weatherhead Center for International Affairs (WCFIA), and a Fellow of the Canadian Institute for Advanced Research (CIFAR) in the Boundaries, Membership & Belonging program.

One stream of Professor Nunn’s research focuses on the historical and dynamic process of economic development. In particular, he has studied the factors that shape differences in the evolution of institutions and cultures across societies. He has published research that studies the historical process of a wide range of factors that are crucial for economic development, including distrust, gender norms, religiosity, norms of rule-following, conflict, immigration, state formation, and support for democracy.

Another stream of Professor Nunn’s research examines economic development in a contemporary context. He has published research examining the effects of Fair Trade certification, CIA interventions during the Cold War, foreign aid, school construction, and trade policy. He is particularly interested in the importance of the local context (e.g., social structures, traditions, and cultures) for the effectiveness of development policy and in understanding how policy can be optimally designed given the local environment. Specifically, he has studied the relationship between marriage customs and female education, generalized trust and political turnover, the organization of the extended family (lineage) and conflict, and traditional local political systems and support for democracy.

His current research interests lie in better understanding the importance of local culture and context for economic policies, particularly in developing countries.

Zoom meeting ID: 992 6583 2638
Join Zoom meeting: https://mit.zoom.us/nathannunn

Please contact sbergen@mit.edu to register for this meeting.

YouTube livestream: https://youtu.be/Mjb6Mq-I1ZE

The last half century has seen the adoption of democratic institutions in much of the developing world. However, the conditions under which de jure democratization leads to the representation of historically disadvantaged groups remains debated as do the implications of descriptive representation for policy inclusion. Using detailed administrative and survey data from Nepal, we examine political selection in a new democracy, the implications for policy inclusion and the role of conflict in affecting political transformation. I situate these findings in the context of the political economy literature mapping institutional choice to power and inequality.

About the speaker: Rohini Pande is the Henry J. Heinz II Professor of Economics and Director of the Economic Growth Center, Yale University. She is a co-editor of American Economic Review: Insights.

Pande’s research is largely focused on how formal and informal institutions shape power relationships and patterns of economic and political advantage in society, particularly in developing countries. She is interested the role of public policy in providing the poor and disadvantaged political and economic power, and how notions of economic justice and human rights can help justify and enable such change. Her most recent work focuses on testing innovative ways to make the state more accountable to its citizens, such as strengthening women’s economic and political opportunities, ensuring that environmental regulations reduce harmful emissions, and providing citizens effective means to voice their demand for state services. In 2018, Pande received the Carolyn Bell Shaw Award from the American Economic Association for promoting the success of women in the economics profession. She is the co-chair of the Political Economy and Government Group at Jameel Poverty Action Lab (J-PAL), a Board member of Bureau of Research on Economic Development (BREAD) and a former co-editor of The Review of Economics and Statistics. Before coming to Yale, Pande was the Rafik Harriri Professor of International Political Economy at Harvard Kennedy School, where she co-founded Evidence for Policy Design.

Pande received a PhD in economics from London School of Economics, a BA/MA in Philosophy, Politics and Economics from Oxford University and a BA in Economics from Delhi University.

Reception to follow.