Scientific Argument with Supervised Learning

Jeffrey W Lockhart, Abigail Z Jacobs

Keywords: ML for Discovery, Computer Vision, Neuroimaging, Sex and Gender, Supervised Learning, Classification, Ethics

TL;DR: Supervised learning, frequently used in the study of differences between human groups, can obscure and legitimize potentially harmful assumptions; we point to some strategies to avoid these pitfalls.

Abstract: The use of machine learning (ML) for scientific discovery has enabled data-driven approaches to new and old questions alike. We argue that scientific arguments based on algorithms for discovery hold the potential to reinforce existing assumptions about phenomena, under the guise of testing them. Using examples from image-based biological classification, we show how scientific arguments using supervised learning can contribute to unintended, unrealistic, or under-evidenced claims.

Track: Attention Track