Proposed Guidelines for the Responsible Use of Explainable Machine Learning

Explainable machine learning (ML) enables human learning from ML, human\nappeal of automated model decisions, regulatory compliance, and security audits\nof ML models. Explainable ML (i.e. explainable artificial intelligence or XAI)\nhas been implemented in numerous open source and commercial packages and\nexplainable ML is also an important, mandatory, or embedded aspect of\ncommercial predictive modeling in industries like financial services. However,\nlike many technologies, explainable ML can be misused, particularly as a faulty\nsafeguard for harmful black-boxes, e.g. fairwashing or scaffolding, and for\nother malevolent purposes like stealing models and sensitive training data. To\npromote best-practice discussions for this already in-flight technology, this\nshort text presents internal definitions and a few examples before covering the\nproposed guidelines. This text concludes with a seemingly natural argument for\nthe use of interpretable models and explanatory, debugging, and disparate\nimpact testing methods in life- or mission-critical ML systems.\n