Ask Ashish Bansal what the top three machine learning algorithms have in common, without hesitation he says “extremely explainable”
(Source – Kaggle survey)
Being a Data Science Director in the heavily regulated finance industry, Ashish is perhaps more sensitive than most regarding the importance of having trustworthy interpretable AI.
We agree with Ashish on the relevance of explainability and believe it will be of growing importance even in industries with far fewer regulatory constraints. The desire for accelerated adoption, strengthened robustness and enhanced insights will drive XAI.
We also agree that logistic regression, decision trees and random forests are relatively more explainable than the zoo of neural networks. However, keep in mind that even with those simpler techniques it is not always obvious how to produce the type of explanation that a particular stakeholder would most value.
Given the excitement over neural networks it will be interesting to see where they rank in next years survey.