Brian Moynihan To Figure Out AI Before Using It

In other words, don’t count on a robot takeover of BofA any time soon.

Tech savvy is perhaps not Bank of America’s most noted characteristic. So while it makes a futile effort to win over the hearts and minds of the young people with a somewhat-less-than-even-halfhearted bit of lip service vis-à-vis gun control, it’s also driving them away by refusing to let them trade these newfangled "cryptocurrencies" they all seem so into these days. But give Brian Moynihan & co. this much: They are committed to learning. And so before deploying that latest doubled-edged sword, artificial intelligence, to reject customer fraud claims in a cool new 21st-century way, it’s gonna do a little homework to prevent it from blowing up in its face a la Countrywide.


The bank currently uses analytic models to help employees detect fraud and, for the past few months, has been studying how artificial intelligence could further improve the detection rate, he said.

If a customer reports a fraudulent charge, an advanced AI system could one day analyze large datasets and provide an employee with a specific judgment about whether the charge was indeed fraudulent, based on the customer’s past purchase behaviors and other data….

But the problem with advanced artificial intelligence systems, such as deep learning, is that they are not well understood…. “We want to understand how the decision is made, so that we can stand behind it and say that we’re not disfavoring someone,” he said.

Bank of America Confronts AI’s ‘Black Box’ With Fraud Detection Effort [WSJ]