1

Dropping Just a Handful of Preferences Can Change Top Large Language Model Rankings
We present a method for auditing the robustness of LLM ranking systems to worst-case data-dropping; we find that dropping just 0.02% of human (and AI) preferences can change the top-ranked models on Chatbot Arena.
Jenny Y. Huang*, Yunyi Shen*, Dennis Wei, Tamara Broderick
PDF Code
Moving towards a more equal world, one ride at a time: Studying Public Transportation Initiatives using interpretable causal inference
Our work identifies impact of subsidy in transit fares on ridership behaviour, and gives insight on how to formulate policy to make cities more equitable through policy intervention that increases transit access.
Jenny Huang*, Gaurav Parikh*, Albert Sun*, Lesia Semenova, and Cynthia Rudin
PDF Code
Moving towards a more equal world, one ride at a time: Studying Public Transportation Initiatives using interpretable causal inference
The goal of low-income fare subsidy programs is to increase equitable access to public transit, and in doing so, increase access to …
Gaurav Parikh*, Jenny Huang*, Albert Sun*, Lesia Semenova, and Cynthia Rudin
PDF Code
Dropping Just a Handful of Preferences Can Change Top Large Language Model Rankings
We present a method for auditing the robustness of LLM ranking systems to worst-case data-dropping; we find that dropping just 0.02% of human (and AI) preferences can change the top-ranked models on Chatbot Arena.
Jenny Y. Huang*, Yunyi Shen*, Dennis Wei, Tamara Broderick
PDF Code