Accountable Computer Systems

We rely on computer systems to store, process, and transmit practically all data whose reliability, accuracy, authenticity, privacy, security, and integrity are vital and regulated. Applications in human resources, medicine, education, finance, and public surveillance use machine learning models to make critical decisions that directly affect people. However, the systems on which these applications run are opaque; we rarely know what decisions are being made, how they are being made, why they are being made, and the degree of certainty any piece of software has regarding these decisions. This lack of transparency erodes public trust and deprives people of their agency. Solutions to this problem lie at the intersection of technology, recordkeeping and preservation, law, public policy, and business. Yet, few individuals understand the language, concepts, constraints, requirements, and possibilities in more than one of these fields, let alone all. This course will bring together students from a diverse set of backgrounds who will learn from an equally diverse group of faculty and outside experts in law, computer science, public policy, artificial intelligence, digital records management and preservation, philosophy, and machine learning how to identify real problems that might require technical or partially-technical solutions, the language in which to communicate between multiple disciplines, and the possible approaches for addressing the most pressing challenges. We invite graduate students to consider the complex societal impact of computer systems. Students will form interdisciplinary teams to undertake a project of their choice, reimagining technology to reduce its negative impact based on technical, ethical, socio-economic, and legal considerations.