Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead

Black box machine learning models are currently being used for high stakesdecision-making throughout society, causing problems throughout healthcare,criminal justice, and in other domains. People have hoped that creating methodsfor explaining these black box models will alleviate some of these problems,but trying to explain black box models, rather than creating modelsthat are interpretable in the first place, is likely to perpetuate badpractices and can potentially cause catastrophic harm to society. There is away forward – it is to design models that are inherently interpretable. Thismanuscript clarifies the chasm between explaining black boxes and usinginherently interpretable models, outlines several key reasons why explainableblack boxes should be avoided in high-stakes decisions, identifies challengesto interpretable machine learning, and provides several example applicationswhere interpretable models could potentially replace black box models incriminal justice, healthcare, and computer vision.

Further reading#

• Access Paper in arXiv.org