Transparent ML for Enterprise Decisions — Intro
Note: this is Part I in a series of articles on transparent machine learning models, click here for Part II — Linear Models, Part III — Rule Sets and Part IV — Scorecards.
The Case for Transparency in Enterprise Decisioning
Remember the giddy excitement when AlphaGo first beats the reining Go champion Lee Sedol? AlphaGo is described as being “creative”, showing “human-like abilities” and “thinking differently”.
Five years later, there are now hundreds of videos and articles analyzing the Go challenge — including dozens on the infamous “Move 37” alone — including a full-length award-winning documentary where the DeepMind scientists discuss how AlphaGo “reasoned”.
Can you imagine similar commentary in an insurance company, at the headquarters of a large retailer or perhaps inside the mortgage division of a bank?
“Amazing, our new Machine Learning model just approved that loan that our old policies would have rejected! Such creativity! What potential can it possibly observe in that mortgage application that we can’t detect ourselves? Can’t wait to see whether the customer pays us back or…
