This paper studies how AI persuades a decision-maker who must attribute the AI’s conflicting recommendation to one of two sources of disagreement: attention differences, where the AI detects features the decision-maker missed, and comprehension differences, where the AI and the decision-maker interpret observed features differently. We show that AI is more effective in persuading the decision-maker when the disagreement is attributed to attention differences rather than comprehension differences. We also show that the AI’s interpretability shapes how the decision-maker attributes the sources of disagreement and, in turn, whether they follow the AI’s recommendation. Our main result is the paradox of interpretable AI: making AI uninterpretable can enhance persuasion and, in the presence of career concerns, improve decision accuracy.