Ziwei Gu

We propose a method to extract the style dimensions people actually care about just by sampling the same prompt many times and running PCA on the model's activations, with no training needed.

Large document collections can be overwhelming. We investigate ways to enable comparative close reading at scale by ordering, aligning, and highlighting analogous sentence segments.

Through an eye-tracking study, we uncover unique gaze patterns when people read with GP-TSM, an LLM-powered reading support tool, and explain how it shapes the reading experience.

AI summaries can be helpful but also unreliable. We design an LLM-powered technique that enhances reading efficiency while making readers resilient to errors in summaries.

LLMs can quickly generate many responses to a single prompt. Each might be better in some way. We explore how to present many LLM responses at once to support sensemaking at scale.

Interaction logs are useful but can be extremely complex. We break down event logs into goal-directed segments to make it easier to understand user workflow.

How do core design elements of debiasing systems shape how people reason about biases? We find distinctive sensemaking patterns for different systems through a think-aloud study.

We present Silva, an interactive tool that utilizes a causal graph linked with quantitative metrics to help people find and reason about sources of biases in datasets and machine learning models.