Good morning, Madam Chair, and members of Parliament and fellow panellists. Thank you for the invitation to appear.
The study of differential outcomes in IRCC decisions is definitely a complex and expansive topic. I'm happy to take questions on that generally, but for purposes of my opening remarks, I'm going to focus on advanced analytics as part of AI, artificial intelligence, solutions.
My professional career has been dedicated to the practice and study of immigration, citizenship and refugee law, or immigration law for brevity. For the past near 25 years, I have accompanied applicants in the journey from the application stage to the Supreme Court of Canada and back, working to advocate preservation of the foundational legal rights of equality, fairness and individualization in immigration law and policy.
AI engages all of these issues. In fact, for some who, like me, are grappling with the enormity of the implications surrounding the expanded use of AI solutions, we can find ourselves equally in awe and in dread. Simply put, the awe part is the exciting and transformative possibilities, such as expedited processing, minimizing officer error or bias, enhanced user access and better information gathering, among other potential gains. However, part of the dread arises from concerns AI will only perpetuate existing racial, socio-economic and political divides and discrimination that are so entrenched in our society.
AI, as we all know, is already being used to triage applications, but the potential is so much more and the plans are far greater than triaging.
The hope is that if AI is applied properly, we can avoid the missteps we've seen and witnessed domestically and internationally when AI has reproduced bias, such as in the Supreme Court of Canada case of Ewert. Other examples internationally include the risk classification assessment tool used for detention in the United States or the iBorderCTRL lie detector used by the European Union at borders, or even the New Zealand technology to identify potential overstayers. These are just a small number of examples of AI use gone wrong. At its core, until we are all required by law to be on a relatable informational plane, applicants, the public and many stakeholders remain at a disadvantage in understanding how parts of their lives may be reordered by AI.
I remain optimistic that improvement is possible. Recommendations this committee has already heard with respect to an ombudsperson and enhanced IRCC training, as set out in the Pollara report, are potential important steps, but even more will be required in the presence of disruptive technology to avoid potential discriminatory consequences.
As set out in the 10 recommendations in our brief, efforts to transform immigration delivery must include legislative measures to be put in place for transparent, explainable and equitable AI governance, recognizing the technology is not neutral and that historical data values and norms propel AI. Training AI programmers, analysts and thought leaders, as one example, must not only be reflective and responsive to vulnerable persons and groups, but must also be required by law to ensure diversity and inclusion balances are maintained for those who train and drive the technology. An algorithm charter like that in New Zealand, external audits with enforcement powers, and mandatory external consultations are steps that have been recognized internationally as effective and essential for proper AI governance.
We cannot look only at where the use of the technology is today. The need to work with and leverage all stakeholders is acute, given the rapidly evolving challenges ahead. IRCC is staffed by many hard-working, well-intentioned individuals who want to make a positive change. Our council and the academic and AI community in Canada are also well positioned to make a meaningful contribution to IRCC's use of AI.
If we put in the collective work now, pressing for a strong and modern legislative framework predicated on collaboration, oversight, transparency, and responsible implementation, we have the potential to be a world leader. There is much work to do, but this is the time to reimagine, revolutionize and reorder Canadian immigration decision-making, built on a strong legal and technological foundation that is grounded in legislation, a foundation undisturbed no matter the international and technological pressure that may be on the horizon.
Thank you.