How AI could be a useful tool in refugee hearings

How can artificial intelligence (AI) be used to assist judges who are tasked with making life-changing legal decisions on if asylum seekers can stay in Canada?

Toronto Metropolitan University’s (TMU) professor Hilary Evans Cameron of the Lincoln Alexander School of Law, an expert in refugee law, teamed up with marketing professor Avi Goldfarb, the University of Toronto’s Rotman Chair in Artificial Intelligence and Healthcare, to examine how prediction technology such as AI could be useful in such hearings.      

At the heart of professor Evans Cameron’s legal research is the question of what is the wrong mistake: sending a legitimate refugee home to face danger and persecution, or letting someone with a more questionable claim to asylum stay? The first is the wrong mistake of the two potential errors, she says. Professor Goldfarb is interested in how AI can be used to make predictions and support human judgments. The two began to collaborate after reading each other’s books—professor Evans Cameron’s on refugee law and professor Goldfarb’s on AI and prediction—and realizing there were synergies in their work.  

Their joint research assessing the potential ways to apply AI in refugee law led to the idea of employing it as a tool to assist judges in reading and assessing the voluminous country reports used to detail the potential threats facing refugees in their countries of origin, such as how police might respond to a request for help. These reports can be very dense and difficult to analyze, professor Evans Cameron notes. “Our minds aren’t really made to process 500 pages,” said professor Evans Cameron.  

Such an AI tool would not replace human judgment, but it could offer an assessment of the contents and the quality of information in the report with regard to the risks, as the duo suggests in their research. It can be easy for humans to gloss over uncertainties, but this tool would highlight any doubts that the judge should consider when deciding the case. Professor Goldfarb notes that you could use AI to assess the likelihood that a claimant’s prediction about the dangers faced when returning to the country of origin, such as if police would respond to a request by them for help, is accurate.   

Migration laws are in flux around the world, professor Evans explains. Countries are taking actions to limit influxes of refugees, such as the U.S. limiting migration through their southern border. She anticipates that international conventions and countries’ individual laws surrounding asylum seekers will change dramatically over the next decade. “There is a very real chance that in the next 10 years, we will no longer have an international refugee protection regime,” she said. Professor Evans Cameron adds that if changes come to Canada’s refugee regime, there could be the opportunity for the law to recognize that denying protection is the wrong mistake. If so, our system could benefit from using AI in the way described by professors Evans Cameron and Goldfarb. 

There is a very real chance that in the next 10 years, we will no longer have an international refugee protection regime. 

Read more in “Artificial intelligence for a reduction of false denials in refugee claims (external link, opens in new window) ” in the Journal of Refugee Studies.