Responsibly Training Foundation Models: Actualizing Ethical Principles for Curating Large-Scale Training Datasets in the Era of Massive AI Models

Morgan Klaus Scheuerman

Dora Zhao*

Jerone T. A. Andrews

Abeba Birhane

Q. Vera Liao*

Georgia Panagiotidou*

Pooja Chitre*

Kathleen Pine

Shawn Walker*

Jieyu Zhao*

Alice Xiang

* External authors

ACM SIGCHI

2025

Abstract

AI technologies have become ubiquitous, influencing domains from healthcare to finance and permeating our daily lives. Concerns about the values underlying the creation and use of datasets to develop AI technologies are growing. Current dataset practices often disregard critical ethical issues, despite the fact that data represents and impacts real people. While progress has been made in establishing best practices for curating smaller datasets in a more ethical fashion, the unprecedented scale of training data in the era foundation models presents unique hurdles for which AI researchers and practitioners must now face. This workshop aims to unite interdisciplinary researchers and practitioners in an effort to identify the challenges unique to curating datasets for large scale foundation models—and then begin to ideate best practices for tackling those challenges. Drawing from CSCW’s tradition of interdisciplinary exchange, our aim is to cultivate a diverse community of researchers and practitioners interested in defining the future of ethical responsibility in the composition, process, and release of large-scale datasets for foundation model training. We will disseminate the outcomes of this workshop to the HCI community and beyond by developing a conceptual framework of both the
challenges and potential solutions associated specifically with curating datasets for foundation models.