Department of Homeland Security pilots new Artificial Intelligence program for immigration process

The DHS is piloting AI to train officers to review applicants for refugee states in the U.S., however, the machine would not make asylum decisions, according to Secretary of Homeland Security Alejandro Mayorkas. AFP

NEW YORK CITY - As artificial intelligence continues to become a regular part in many aspects of life, the Department of Homeland Security is looking for new ways to integrate the tool into a "labor-intensive" part of the immigration process, Secretary of State Alejandro Mayorkas told reporters in San Francisco.

Concretely, the Department is piloting AI to train officers to review applicants for refugee status in the United States. In this pilot, DHS is training machines to act like refugees so officers can practice interviewing them, assignments that typically involve senior personnel.

Mayorkas also said that the department planned to develop an interactive app to supplement its training of immigration officers, drawing on so-called generative AI that creates novel content based on past data.

"Refugee applicants, given the trauma that they have endured, are reticent to be forthcoming in describing that trauma," Mayorkas said. "So we're teaching the machine to be reticent as well" and to adopt other "characteristics" of applicants.

Despite this training, AI will not make immigration decisions, according to Reuters. The AI will know country-specific conditions and other information to help officers.

Among more "advanced" deployments of AI, Mayorkas said the department has worked to spot anomalies when commercial trucks and passenger vehicles make border crossings. The goal, he said, is to help the department detect smuggling attempts for bringing fentanyl and other contraband into the United States.

This new pilot is part of the many efforts by government agencies to reduce costs and improve performance through AI, according to Reuters. Nevertheless, this process has been met with many challenges, including issues with translation.

Names translated as months of the year, incorrect time frames and mixed-up pronouns are just some of the issues AI-driven translation apps have so far caused in the U.S. asylum system.

"The machines themselves are not operating with even a fraction of the quality they need to be able to do case work that's acceptable for someone in a high-stakes situation," said Ariel Koren, founder of Respond Crisis Translation, a global collective that has translated more than 13,000 asylum applications.