Ethical and Responsible AI as a Means of Diagnosing and Eliminating Bias

Abstract

AI and Machine Learning (ML) are powerful tools that allow us to analyze and derive knowledge from information that may not be accessible using more traditional methods. AI/ML opens the gateway to exploring more complex relationships in both our data and in our practices as a community of scientists. However, the results of any ML model must be examined to ensure that they are valid and beneficial; otherwise the practitioners may act on false or misleading results. The NASA Framework for the Ethical Use of AI identifies principles and practices that are fundamental to Ethical AI. There are many ways that Ethical AI can be leveraged to improve equity and fairness in our field. For example, there are practices in machine learning that can be used to clearly diagnose the factors behind human decisions, allowing us to pinpoint the presence and sources of bias. This can become a roadmap for ensuring fairness in future decisions. This presentation will review strategies for implementing Ethical/Responsible AI and discuss how they can be used to advance Diversity, Equity and Inclusion.