The growing use of opaque artificial intelligence (AI)-driven methods by banks is creating a problem for regulators, Benoît Cœuré said today (August 19).

The head of the Bank for International Settlements’ innovation hub said a well-known problem in AI was that it could be difficult for a human to understand how an AI model arrived at its conclusions – often called the “black box” problem.

Cœuré drew a parallel to the global financial crisis, when the “complexity and opacity of internal models