At a press conference held in Vancouver, Immigration Minister Sean Fraser asserted that by increasing the use of technology, advanced analytics, & AI in application processing, the government has done a couple of essential things in recent months.

He further stated about the TRV applications for individuals waiting for permanent residency under family class sponsorship programs. However, the increased use of AI in application processing has had an impact across lines of businesses by IRCC. 

In the forthcoming years, Canada will welcome the most immigrants ever. In 2022, IRCC made over five million strategic decisions on applications for all business lines. This number was double as compared to that in 2021. 

IRCC has rapidly made a shift towards digitization & modernization of Canada’s immigration system. This includes increased use of advanced analytics & AI in application processing. For instance, the minister asserted that this streamlining has led to the processing of over 98% of spousal TRV applications. Moreover, new candidates can now expect a processing time of just 30 days. 

IRCC says digitizing the procedure will help meet the needs of clients. Further, the department says that leveraging technology and using AI in application processing will free the officers to focus more on complex applications. 

IRCC utilizes AI in some aspects of processing, like: 

  • Automating positive eligibility determinations 
  • Distributing applications between officers depending on application characteristics
  • Workload distribution 
  • Recognizing applications that might need extra verification
  • Triaging client emails to get fast replies 
  • Assessing biometrics 

 

There are no algorithms that IRCC uses to reject or accept a candidate. Instead, applicants are approved or rejected based on a visa permit on an AI-generated decision. 

 

What Are The Concerns? 

Many people worry that using AI in application processing will lead to bias & general lack of explanation for how decisions are made. 

For instance, the Treasury Board Directive on Automated Decision-Making asserts that basing an algorithm on historic data amplifies race, caste, gender, & other inequalities. Moreover, it has facial recognition software that might not work equally well for all skin tones & genders. 

In addition, depending too heavily on AI for application processing can lead to a lack of clarity in decision-making. The Directive further says that the federal government must be able to explain how decisions are made. Moreover, applicants denied services have the right to get a clear and understandable explanation from the government. The explanation must go beyond insinuating that it was a decision given by the computer/ AI. 

Many concerns have also been raised around the usage of ‘Chinook’, a Microsoft Excel Tool. It is used to simplify the virtual representation of a client’s information. In addition, IRCC uses it to assess temporary visas, work permits, study permits, etc. 

The department asserts that ‘Chinook’ does not use AI for making decisions.  

 

How IRCC Develops Algorithms? 

Responding to the concerns, IRCC says it follows a Directive to ensure equal rights & prevent discrimination. It further states that it follows principles of transparency, legality, accountability, and procedural equality to define how decisions are made and what explanations are offered to impacted people.   

Whenever a new algorithm is proposed, it should be approved by the AIA, i.e., Algorithm Impact Assessment. The AIA is a crucial assessment tool & a part of the Directive. This tool measures risk areas, mitigation, & effect of the algorithms. Moreover, IRCC asserts they were one of the first government departments to use AIA.  

The department also says it needs assessments of:

  • Effect of algorithms
  • Quality assurance measures for information & algorithm 
  • Proactive disclosures of how and where algorithms are used

 

IRCC states that the rules used to support the decision-making process are reviewed by professionals & experienced officers. This ensures that the process aligns with the eligibility criteria outlined in the IRPA, i.e., Immigration & Refugee Protection Act. 

Moreover, consistent monitoring & quality assurance measures must be in place to ensure that these models continue to perform as they used to, thus identifying & mitigating any unforeseen negative impacts. 

The immigration minister remains optimistic about using AI in application processing & advanced analytics by IRCC. He asserted that the technology adopted a few years ago enables the department to look across several factors on an applicant’s application to help determine the likelihood of being eligible for permanent residency. 

He further states that this has resulted in increased productivity and continues to stress upon the fact that AI does not make any vital decisions.