Developing A Conceptual Framework for Analyzing People in Unstructured Data

Published: 23 Oct 2023, Last Modified: 28 Nov 2023SoLaR PosterEveryoneRevisionsBibTeX
Keywords: data-centric AI, responsible AI, sociotechnical evaluation
TL;DR: We propose a framework to guide analysis of human representation in unstructured data and identify downstream risks.
Abstract: Unstructured data used in foundation model development is a challenge for systematic analyses to make data use and documentation decisions. From a Responsible AI perspective, these decisions often rely upon understanding how people are represented in data. We propose a framework to guide analysis of human representation in unstructured data and identify downstream risks.
Submission Number: 41
Loading