Computer says no: combating bias in machine learning models

Machine learning is at the root of many recent advances in credit risk management. Techniques capable of making lending decisions far more rapidly, accurately (and cheaply) than a human are common at most large banks and credit card issuers. Self-learning algorithms are also at the front line of the battle against customer fraud. But what if those algos were just as fallible and flawed as the human loan managers who went before them?

As well as growing issues over the lack of ready