Dissertation Title: Public Interest Computing: A Pluralistic Design Language Foundation for Societal-Machine Alignment

Abstract

As algorithmic systems, including artificial intelligence (AI), become increasingly prevalent in societal decision-making, ensuring their alignment with public interests is crucial. The reciprocal relationship between societal norms and these technologies—where our values shape their design, and they, in turn, influence human interactions—calls for a fundamental reconceptualization of AI ethics at a structural level. This research introduces Public Interest Computing, a Pluralistic Design Language that creates a novel space for value-sensitive algorithmic ecosystems, fostering societal-machine alignment. 

We begin by developing a comprehensive theory of Public Interest Computing, rooted in the capability approach. Building on this theoretical foundation, we collaborate with communities to co-design real-world systems and inform policy, each tackling a distinct aspect of societal-machine alignment. These Public Interest Computing systems demonstrate versatility in addressing societal-machine alignment challenges. Farm Pulse exemplifies substantive fairness in data-driven decision-making, enhancing access and equity to climate change adaptation for at-risk farmers. Boomerang exhibits incentive-compatible alignment, promoting equitable designs in AI data markets. The Prototype Tasks System illustrates computationally mediated cognitive alignment, improving fairness within these markets. The Beyond Boundaries framework enables environmental alignment, providing a platform for public discourse on climate change. Complementing these implementations, our analysis using Gobo focuses on value alignment, investigating ways to increase human agency in interactions with invisible algorithms on online platforms. We conclude by discussing the broader research impact of Public Interest Computing, including its role in an AI Policy Initiative focused on data-driven humanitarian mapping, and its potential to drive global reflection and local action. 

As our relationship with technology continues to evolve, Public Interest Computing and its Pluralistic Design Language serve as a compass, guiding us towards a future where societal values and algorithmic ecosystems are inherently intertwined. This endeavor is not a destination but an ongoing process of reflection, adaptation, and collaboration, ensuring that as our relationship with technology progresses, so too does our commitment to aligning it with the greater good.

Committee

Danielle Wood, Associate Professor of Media Arts and Sciences, MIT Media Lab and AeroAstro
Julie Shah, Department Head, AeroAstro, H.N. Slater Professor of AeroAstro, Associate Dean, Social and Ethical Responsibilities of Computing, Schwarzman College of Computing
Sally Haslanger, Ford Professor of Philosophy, MIT Department of Linguistics and Philosophy
Bish Sanyal, Ford International Professor of Urban Development and Planning MIT Department of Urban Studies and Planning
Nabeel Gillani, Assistant Professor of Design and Data Analysis, Northeastern University