Explainability covers the degree to which a system's reasoning can be understood. In practice for legal applications it means showing which sources were retrieved and which passages support each statement, rather than exposing model internals.
Full interpretability of large language models remains a research problem, but source-level explanation is achievable and is usually what legal users actually need.
Alternative Names:
Interpretability, Explainable AI
Why it Matters?
Lawyers do not need to understand model weights, but they cannot rely on an assertion without knowing what supports it. Source-level explanation is the practical form of explainability that makes verification possible and makes AI-assisted analysis defensible when challenged. Vendors describing explainability without providing traceable citations are usually describing something other than what a litigator needs.
Frequently Confused with
Related terms
Frequently asked questions
Does explainability mean understanding how the model works internally?
Why does explainability matter in litigation?
