DemoPricing

What is AI Explainability in Regulated Decisions?

Explainability for a regulated decision is the ability to say why that decision came out as it did: which rule applied, and which facts met or failed it. It is decision-specific, not a general account of how a model weighs its inputs.

Rule-level explanations

When a decision is taken by sealed rules, the explanation is the rule itself: the condition that was met, the facts that met it, and the sentence of the policy it comes from. It is exact, and it is the same every time the same facts are given.

Where models are involved

A model's output, a risk score for example, can be one of the facts a rule reads. The rule then explains what was done with the score; explaining the score itself remains a question about the model.

Example

Scenario

A mortgage application is declined.

Without governance

The applicant is told that income and credit history were the most important factors overall, which says nothing about their own case.

With NOMOS

The record says: rule MORT-012 applied; monthly debt obligations of AED 8,400 exceeded 50% of verified income of AED 16,500; the rule cites section 4.2 of the credit policy.

Frequently asked questions

Related concepts

See it in practice

Upload a policy, confirm its rules, seal them: any system can then check an action against them.

Open Studio →