AI translation is jeopardizing Afghan asylum claims

June 20, 2020

A Pashto-speaking refugee's asylum claim was denied due to a discrepancy between oral and written recount of an event. The error? An automated translation tool misinterpreted the pronouns in her written statement, swapping 'I' for 'we'. This underscores the need for responsible AI governance and the importance of trustworthy AI in our society.

Join us at HISPI Project Cerebellum to help establish guardrails for AI and prevent such incidents through the Measure function of our TAIM platform.
JOIN US

Matched TAIM controls

Suggested mapping from embedding similarity (not a formal assessment). Browse all TAIM controls

MEASURE 2.6 — similarity 0.631, rank 1. TAIM detail and related incidents →
MAP 1.6 — similarity 0.628, rank 2. TAIM detail and related incidents →
MAP 4.1 — similarity 0.612, rank 3. TAIM detail and related incidents →

Alleged deployer: us-citizenship-and-immigration-services
Alleged developer: unknown
Alleged harmed parties: anonymous-pashto-speaking-refugee, pashto-speaking-asylum-seekers, dari-speaking-asylum-seekers

Source

Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/532

Data source

Incident data is from the AI Incident Database (AIID).

When citing the database as a whole, please use:

McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.

Pre-print on arXiv · Database snapshots & citation guide

We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.