All AI projects

Humanitarian AI case study

DigitalAAP: AI for Refugee Feedback and Accountability

DigitalAAP is a voice-first accountability platform that helps humanitarian teams transcribe, classify, route, and review feedback from affected people without removing human responsibility.

Organization
UNHCR
Period
2023
Role
Project lead

The challenge

Feedback from displaced people often arrives as speech, across languages and dialects, through channels that do not create structured records. Manual review limits how quickly concerns can reach the team able to act.

The approach

  • Use accessible voice channels rather than requiring an app, written literacy, or a data plan.
  • Apply speech-to-text and classification to prepare feedback for triage while retaining human review for sensitive or ambiguous cases.
  • Build privacy, consent, escalation, and accountable ownership into the workflow instead of treating them as a final compliance review.

Measured outcomes

  • Selected for the third cohort of the UN Global Pulse Accelerator Programme.
  • Published in the UN Innovation Network library as an example of refugee feedback through AI.
  • Created a reusable operating pattern for multilingual feedback intake, classification, escalation, and management reporting.

Governance and safeguards

  • Human review remains in the path for sensitive or uncertain feedback.
  • The system supports triage and routing; it does not make protection or assistance decisions.
  • Access and data use are constrained by humanitarian data-protection requirements.

Evidence and sources

The links below distinguish independently published evidence from first-person programme records.

Related guidance