Explainability

Explainability (XAI)

Intro
Explainability addresses the tension between model complexity and accountability.

1. What is it?
Explainability refers to the ability to understand how an AI system produces outcomes.

2. What problem does it claim to solve?
It aims to make AI decisions transparent enough for oversight and challenge.

3. Where does it typically appear in organisations?
Regulated decision processes, financial reporting support and risk models.

4. What can go wrong if it is misunderstood or misused?
False assumptions about explainability may lead to overreliance on opaque models.

5. Who is accountable, and what oversight is required?
Management must ensure explainability is sufficient for the decision context; boards must assess acceptability.