In this talk, Perez explores to what extent language models can be evaluated through processes generated by the models themselves, presenting recent research that demonstrates successful results, as well as enabling the discovery of novel behaviours within the models.

In this talk, Goldfarb explores how artificial intelligence (AI) presents opportunities and threats in ways that are both extraordinary and unexpected. Identifying AI’s ability to decouple prediction from other aspects of decision-making as a key to its transformative impact, Goldfarb contends that it will require the invention of new ways of operating—many of which remain undiscovered—in order to truly unleash AI’s innovative potential.

Our weekly SRI Seminar Series welcomes Jon Lindsay, an associate professor at the School of Cybersecurity and Privacy at the Georgia Institute of Technology, with a joint appointment in the Sam Nunn School of International Affairs and a courtesy appointment in the School of Public Policy.

In this talk, Penney will explore how rapid technological developments are leading to the automation of legal enforcement, generating serious risks for privacy and human rights. Amidst a lack of guidance for lawmakers and policymakers grappling with these issues, Penney will consider theoretical and empirical research on the social impacts of these technologies, and the shortcomings of typical solutions such as regulation and human oversight.

In this talk, Evans will present recent work on defining and measuring “truthfulness” in the context of large language models, including their calibration, and their ability to forecast world events. These topics will be considered in relation to the reduction of epistemic harms from AI and the problem of value alignment in the context of artificial general intelligence.

In this session, Raso explores how technologies that administer border and immigration policies construct another equally important, but less explored, subject: state agency. Drawing on a recent example from Canada, Raso demonstrates how digitalization simultaneously concentrates state power while diffusing agency, reflecting on what this means for legal accountability mechanisms and decision-making.

In this talk, Suresh Venkatasubramanian will dissect the Blueprint for an AI Bill of Rights, discuss what its authors were trying to achieve, illustrate the difficult issues of technology and policy that had to be resolved, and expand on where this effort is going.

In this talk, Ariel D. Stern will explore how the rapid innovations of digital technology are changing definitions and approaches towards healthcare by practitioners, institutions, and users, and why these changes require new and innovative approaches towards policy and regulation.

Our weekly SRI Seminar Series welcomes Kobbi Nissim, the McDevitt Chair of Computer Science at Georgetown University, and an affiliate professor at Georgetown Law. Nissim’s research works towards establishing rigorous practices for privacy in computation. He is particularly interested in intersection points between privacy and various disciplines within and outside computer science, including cryptography, machine learning, game theory, complexity theory, algorithmics, statistics, databases, and more recently privacy law and policy.