BIAS Blog 2: Harms in Machine Learning
Published on:
How can Machiene learning be harmful
New Article
Understanding Potential Sources of Harm throughout the Machine Learning Life Cycle
The main purposes of this case study is to educate learners to think about the data as the product of a complex human-driven process, as opposed to a static, objective artifact.This case study provides a framework that identifies seven distinct potential sources of downstream harm in machine learning, spanning data collection, development, and deployment. It explains how such issues arise, how they are relevant to particular applications, and how they motivate different mitigations. After reading this case study, we will think about data as the product of a complex human-driven process, as opposed to static, objective artifact.
We describe how these issues arise, how they are relevant to particular applications, and how they motivate different mitigations.