|
Our firm has led key litigation in Federal Court involving substantive evidence of the working of the Chinook system and the impact that it has on decision making.
There are fundamental challenges that we face in proving the harm caused by these systems and the direct impact that they have on immigration outcomes.
Mehrara v. Canada (Citizenship and Immigration), 2024 FC 1554
As AI and Automated Decision-Making (ADM) tools become more prevalent in administrative law, including Canadian visa processing, the challenge of contesting decisions influenced by these technologies continues to grow. While previous court rulings have touched on the use of the Chinook tool in visa refusals, no decision has fully explored its inner workings - until now.
In Mehrara , we presented evidence on how Chinook spreadsheets function and how they impact decision-making in study permit applications. The key challenge in Judicial Review cases like this is the lack of transparency. Critical evidence, such as how data is processed, what decision-makers actually see, the role of AI versus human input and identifying what is AI generated vs human generated, is controlled by IRCC. Unfortunately, in this case, we couldn't fully establish how the use of Chinook affected procedural fairness, as vital evidence (e.g. deleted spreadsheets) was inaccessible to us.
This case marks an important step in the ongoing legal analysis of how AI and ADMs influence administrative decisions and the framework we need to govern these technologies. I’ll be writing in more about this case in the coming weeks, but ultimately Justice Battista's decision highlighted the limitations of the available evidence and acknowledged that the issue of AI's role in decision-making is far from resolved:
“[64] However, while I have found the record to be complete for the purposes of reviewing this decision based on the nature of the error, the reasons of the Officer, and the current evidence regarding the function of Chinook, this may not be the case in other judicial reviews of applications processed using processing technology, particularly in applications where risk indicators are present. For this reason, the Respondent’s systematic daily deletion of all material generated by processing technology may not reflect best practice. The growing recognition that the exercise of public power must be justified, as articulated in Vavilov and affirmed in Mason, would be hollow without a basic understanding of how the exercise of that power occurs.”
However, our work in this space continues and we welcome cases where applicants believe the use of technology tools has negatively impacted the outcome of their case.
|