AI Trust Patterns

ExplainabilityTrust and Transparency

Multi-Audience Explanation

Explain the same issue differently to the reviewer who resolves it and the person it affects.

Phase
Design · Build
Applies to
End users, content designers, UX designers
Oversight
Human in the loop
Key factor
Rights & data sensitivity
OverviewWhen to useQuestions to askExampleAnti-patternAnti-patterns to pay attention toReferencesRelated

Overview

When an AI flags an issue, more than one person is affected: the reviewer who resolves it and the person affected by the final outcome. Multi-Audience Explanation builds on Accessible Communication by tailoring the same issue for each of them. The reviewer gets evidence, context and tools to resolve it. The affected person gets a respectful, clear explanation with a direct path to human review, without duplicated content or exposed internal mechanics.

When to use

Use when

  • An AI result affects more than one person, or there are several personas in a workflow.

Not needed when

  • The reviewer and the affected person are the same, like someone checking their own AI-drafted email.

Questions to ask

Ask these at kickoff, when design work starts. Tick them off as your team answers.

Example

An AI notices that the income on a rental application doesn't match the payslip.

One flag · Two audiencesDo
Reviewer
Check: income mismatch
Form says $5,200/month; payslip (Sept) shows $4,350. Two sources disagree.
Compare documentsRequest documentOverride with reason
Applicant
We need one more document to finish your application: your most recent payslip.
A member of our team reviews every application. Questions? Contact a person and we'll help.
Upload payslip

Anti-pattern: One screen for both

The reviewer's flag, shown to the applicant as is. It's alarming, it's not actionable, and it exposes internal mechanics.

Applicant viewDon't
Application status: income_mismatch · model confidence 0.41
Your application is on hold.

Anti-patterns to pay attention to

  • The reviewer and the affected people see the same component or copy.
  • Affected people see scores, codes or model terms.
  • Affected people get no explanation and no route to a person.

References

Current guidelines · EU AI Act, GDPR, ISO/IEC 420013 references

For orientation only, not legal advice. Applies to high-risk systems. EU AI Act links open a readable version of the official text published on EUR-Lex; ISO/IEC 42001 is a paid standard.

  • EU AI Act Art. 86 People affected by certain decisions based on high-risk AI outputs have a right to a clear and meaningful explanation of the AI's role and the main elements of the decision.
  • GDPR Art. 22 People have the right not to be subject to solely automated decisions with legal or similarly significant effects, with safeguards including human intervention.
  • ISO/IEC 42001 A.6.2.7 Technical documentation suited to each audience: users, auditors and regulators.