The Center for Responsibility in AI Systems and Experiences (RAISE) advances artificial intelligence that serves the public good, builds trust, and supports open innovation. Our work centers on three aims that define responsible AI in practice.

Applied AI in Underserved Areas
 We study applications of AI in areas underserved by market incentives, especially in the public sector and the physical, life, and social sciences. We aim to improve equitable access and use as the world adapts to the proliferation of AI.

Epistemological Foundations of AI
 We study how AI systems generate and justify claims across pre-training, mid-training, and post-training stages. Our focus is on the epistemological foundations of trust: how errors arise, when abstention is the right choice, how uncertainty should be communicated, and how explanations can support human oversight. By grounding trust in epistemology, we aim to create systems aligned with human values.

Open and Specialized Systems
 Large proprietary models limit transparency, adaptability, and governance. We advance smaller, open-weight, and open-source models that can be adapted and fine-tuned in specialized settings. These models reduce dependency on closed APIs, lower costs, and enable communities to govern AI systems according to their needs.

Together, these aims guide RAISE’s mission to study and deploy AI systems and experiences to serve the public good.