In Europe, if an AI makes a decision about you that affects your life, you have right to know why it made that decision. This “right to know” may be impossible for anything but the simplest systems. The short answer may be that you got declined for the loan because other people like you in subtle but statistically significant ways defaulted on their loans.
The challenge is compounded by the fact that humans don’t really know how we make the decisions that we make. Why did you buy the Ford over the Chevy? You may think you know, but you almost certainly don’t know all of the reasons. Our minds are inscrutable, even to us. There is no reason an AI won’t behave likewise.