Chapter 09 · Sustainability & AISustainability & AI
Explainable AI (XAI)
Definition
Techniques and design approaches that make the behaviour and outputs of AI systems understandable to humans, enabling inspection, debugging and accountability.
References
This source provides part of the technical or institutional basis for the definition.
This source supports the explanation of how the term is applied, measured or governed in practice.
Overview
What it means
Explainability methods range from inherently interpretable models to post-hoc explanations of black-box systems (feature attribution, counterfactuals, example-based explanation). Miller's 2019 synthesis grounded the field in how humans actually explain decisions; the EU's trustworthy-AI framework lists explicability among its requirements.
How it is used
Practitioners apply XAI where reasons matter: credit and risk scoring, screening decisions, scientific use of models, and regulatory contexts requiring 'meaningful information about the logic involved'. Environmental applications include validating that a model uses physically meaningful signals.
Why it matters
An explanation is what turns an output into a decision that can be defended, contested and improved. Where AI touches sustainability compliance or justice, explainability is the difference between a tool and an oracle.