Leveraging AI & ML to uncover unfair practices for those seeking asylum in the U.S.
As a member of the Patrick J. McGovern Foundation (PJMF)’s 2023 Data to Safeguard Human Rights Accelerator program, Mobile Pathways and PJMF’s Data Practice team have partnered to explore data that has exposed unequal access to justice for asylum seekers residing in the U.S.
by Jeffrey O’Brien, Bartlomiej Skorupa, and Ana Luz Ortega Villegas
Amid record-breaking numbers of immigrants seeking safety and justice at the Southern border of the United States, how can we balance making incremental progress against achieving systemic reform of a broken immigration system? Within this challenging context and in pursuit of our mission to develop mobile phone technology to help immigrants navigate the legal immigration process, Mobile Pathways recognized the need to balance swift action while prioritizing data-driven insights to improve the outcomes of asylum seekers in our immigration court system.
Participating in the Patrick J. McGovern Foundation Data Accelerator has been an incredible journey over the past nine months. From exploring a massive federal dataset of immigration court data to building a data pipeline for automated analysis and visualizations, we are immensely proud of the progress we’ve made to see our Asylum Navigator in action. Asylum Navigator is a groundbreaking data insights platform designed to empower immigration attorneys and human rights defenders. It provides instant insights, contextualized data, and easily understandable information regarding asylum seekers’ cases in immigration court.
As our latest analysis sheds light on the harsh reality faced by asylum seekers navigating the U.S. immigration system, the chances of being granted asylum or forced to return to danger and persecution are heavily influenced by a person’s country of origin and current location. This compelling evidence underscores the urgent need for greater attention to asylum seeker rights and fair access to justice.
As a technology nonprofit start-up, we faced numerous challenges on our path to reaching these summary conclusions from our data project. Initially, we fell into the common trap of taking on too much without sufficient resources. Our vision was to provide access to justice for all immigrants in immigration court proceedings while incorporating sophisticated artificial intelligence (AI) and machine learning (ML) practices. Once the work began, we soon realized the daunting challenges ahead. And, as asylum seekers began losing pathways to safety and justice, the pivot to incremental progress became absolutely necessary.
A Time to Pivot
Seeking asylum is an inherent human right–a right that should be respected and protected. It became a legal right in the United States through the Refugee Act of 1980, which established federal protections for individuals fleeing persecution based on their race, religion, nationality, membership in a particular social group, or political opinion.
So why has this legal right to asylum become so controversial? The causes are many, and the situation is worsening. As a result, fair access to asylum is being blocked at an accelerating rate to thousands of people seeking refuge from persecution, violence, and hardship.
Our journey through the Patrick J. McGovern (PJMF) Data Accelerator was characterized by these escalating realities and required nimble actions of our team to adjust accordingly. We embarked on this immigration justice adventure with the hypothesis that AI tools could analyze immigration court data and thereby greatly assist those navigating the asylum process, enabling them to make well-informed decisions. Our aim was to explore the potential of AI in addressing the fundamental human right of fair access to justice for all.
From the outset, we understood the importance of managing scope and not overestimating our abilities. Just like fitting food on a plate, we could only handle a certain amount at once, which is particularly apt advice for a nonprofit technology startup like Mobile Pathways. The PJMF team steered our decision to right-size what we can do now while preparing for even more impact in the future.
Due to a combination of the current immigration backlog, the growing political controversy of asylum, and our own execution capabilities, we made a critical decision to defer the utilization of more sophisticated artificial intelligence and machine learning techniques to later phases. Instead, we adopted an agile approach to use technology tools that our team is already familiar with and develop a visually stunning representation of asylum seeker immigration case data and analysis.
The decision to pivot to what would ultimately become the public launch of Asylum Navigator in September 2023 was absolutely critical. By right-sizing the approach and adopting lean development practices to contextualize immigration court data, our groundbreaking immigration justice tool has already been utilized over 18,000 times to vividly demonstrate the inherent injustice of the asylum process.
Technology Served The Hard Way
The pathway to this success came with many failures and humbling lessons. The road began by analyzing over 100 million rows of information about immigration court proceedings from the public Freedom of Information Act (FOIA) Library. The immigration court data is stable and reliable, as it comes from the monthly extracts published by the Department of Justice.
As encouraged by our Accelerator advisors, we embraced a “Reduce The New” philosophy. Specifically, this meant exploring existing toolsets available in the Salesforce cloud ecosystem–a resource our staff already leverages to inform underserved immigrants of their rights through our MyCamino platform. This “Reduce the New ‘’ approach empowered our nimble data team to quickly develop new capabilities for our partners without starting from scratch, and this meant beginning with the native and powerful Tableau, an interactive data visualization software focused on business intelligence.
Working through the immigration court data dataset and building a working data model underscored the level of complexity in the information and analysis we conducted. As a result, we refined and narrowed our primary audience to immigration attorneys due to their expertise in the complexities of immigration justice and immigration case proceedings. Our efforts were dedicated to studying immigration court case data concerning asylum seekers, a particularly vulnerable population of immigrants in court, and often those who are least likely to secure legal representation and find access to justice.
With a clear roadmap and established end-user, our team meticulously cleaned and normalized the collected immigration court data. This was a considerable undertaking as no public data schema existed, and we had to analyze almost one hundred tables with over a thousand fields, some of which appeared to be duplicates. Data quality and reliability provided additional challenges as critical variables such as gender or date of birth were often incomplete or mislabelled. Through many discussions with our legal experts and partners, plus hours of engineering, we developed a relational model with a snowflake schema and a SQL-based ETL pipeline to house our refined dataset.
With a functional data warehouse in place, we began our analysis with focused attention on three key variables that impact whether asylum seekers in immigration court will be granted asylum:
1. Immigration Court Location
2. Asylum Seeker Nationality
3. Judge
While focused in scope, this initial analysis regarding access to justice provided valuable insights, which can be seen in our “A Lottery Based on Geography” report. The preliminary results were an effective hook in focus group sessions with immigration attorneys. Through these sessions, the attorneys were shocked by the evidence our dashboard provided on inequalities in the judicial system and immediately requested deeper analysis based on motions, requests for a later hearing date, or appeals of the immigration judge’s initial ruling.
With feedback from our first set of users, we took the leap of faith and made our analysis of immigration court data publicly available in September on our website via an interactive Tableau dashboard. The insights gleaned from the data project were sobering, both for our team and our wider immigrant advocate community. Incremental action and advocacy were paramount as we sought to level the playing field and ensure every person seeking asylum gets access to safety and justice.
Clearing The Path To Immigration Justice
The coming months will play a critical role in fortifying the foundation of Mobile Pathway’s data work for the future years. Undoubtedly, our data-driven insights will profoundly contribute to the pursuit of fair access to justice for all asylum seekers. Adoption and awareness of our initial tools are growing quickly, opening new doors for partnerships in Texas and along the border, where increased focus is necessary.
After an intense and fortifying learning journey, our team and partner network are excited to further analyze and uncover more insights within our immigration data. Many of our learnings came through conversations with our cohort members and reflections during technical syncs with the Data Practice experts. The Accelerator has enabled us to turn ideas into reality, transform inaccessible FOIA data into actionable insights for immigration advocates, and generate excitement among our team to iterate and innovate to further leverage data and technology to bring justice to the immigration system.
Together, we can reform the U.S. immigration system, end the unfair asylum lottery, and give every individual the opportunities they deserve for safety, access to justice, and a brighter future. To learn more about our immigration justice findings, please try Asylum Navigator yourself by clicking here.
