Research

The research arm of The AI Clinic prioritizes creating and evaluating new methods and models for rendering language models more approachable for scientists and public interest practitioners. Our work has been published at premiere conferences such as ACM FAccT and SIG-Spatial.

Laboratory-Scale AI: Open-Weight Models Are Competitive with ChatGPT Even Under Low-Resource Conditions

ACM FAccT 2024

Robert Wolfe, Isaac Slaughter, Bin Han, Bingbing Wen, Yiwei Yang, Lucas Rosenblatt, Bernease Herman, Eva Brown, Zening Qu, Nic Weber, and Bill Howe

An analysis of the feasibility of using small, open AI models as opposed to large, closed models.

Mitigating Overconfidence in LLMs

NeurIPS 2024 Workshop on Behavioral ML

Bingbing Wen, Chenjun Xu, Bin Han, Robert Wolfe, Lucy Lu Wang, and Bill Howe

Introduces the Answer-Free Confidence Estimation (AFCE) method for calibrating LLM verbalized confidence assessments.