Signal Preprint examines how research values shape foundation model development

Summary

A preprint analyzes the role that research values play in the development of foundation models. It uses machine learning for medical imaging as its specific case study. The author argues that research values, which carry a normative dimension, shape how researchers make and justify technical decisions. The paper further suggests these values help explain why certain technologies become widely adopted, or conversely remain marginal, in scientific literature and clinical practice. Through this case study in medical imaging, the analysis offers a framework for understanding broader trends in how foundation models are developed.