With AI systems now equaling or exceeding human performance, it’s increasingly important to understand the reasons a prediction is made. This is especially true in banking, where financial stability is at risk if the underlying mechanisms driving market-moving decisions are not well understood and where consumers must be protected from technology-related bias. Cigdem Gurgur explains the limitations and possibilities inherent in XAI and gives examples of AI’s potential to increase accuracy and fairness over the current statistical models guiding credit and lending decisions.