The importance of risk analysis.
Yesterday, a machine learning consultant and I debated what the most crucial up front consideration for a new ML project should be.
At first, 'what are the assets?' seemed the most obvious starting point, but it isn't; unintended consequences are the most crucial thing to bear in mind when consulting for a business.
It's almost always unclear exactly how or why underlying algorithms operate the way they do so you must assess and manage expectations over possible ramifications.
Ask yourself: what are the consequences of a false prediction?
Comments