Interpretability
The ability to understand and explain how an AI model reached its decision.
In Plain English
Interpretability is about transparency: you can look inside an AI's decision-making process and understand which inputs mattered and why. Some models are naturally interpretable—a simple rule-based system or a short decision tree is easy to follow. Others are harder: deep neural networks with millions of parameters can feel like a black box. Improving interpretability often means building tools that reveal which data points the model paid attention to, or creating simpler models that trade some accuracy for clarity. For professionals using AI in regulated industries or making high-stakes decisions, interpretability is critical.
💡Real-World Example
A credit union uses an AI tool to recommend which customers should receive financial literacy coaching. The tool shows a scorecard for each person: 'low savings rate (+15 points), recent late payment (+10 points), long account history (−5 points)'—so loan officers understand the recommendation and can discuss it thoughtfully with the customer.
Related Terms
What did you think of our explanation?