Cat & Mouse - Lessons Learned from Immigration Transformations

In the last 15 years, I have had the opportunity to work on major immigration transformations on two continents and have moved to the other side of the world twice over. What I have come to recognize is that immigration transformation is unique in that it is subject to constantly changing political drivers, must be able to pivot and flex on a moment’s notice, yet the migration of people with various backgrounds and skills are critical to the prosperity of modern societies.

I recently moved from Australia to Canada, and this has given me the opportunity to reflect on what is common around the world when it comes to modernizing immigration systems and what is different; what some countries are doing to increase their attractiveness to immigrants and what others are doing that is not working as well.

This is the third article in my series on Immigration Transformation where I will unpack some of the lessons I have learned working on immigration projects around the world. For my previous articles:

1. Agility is Key
2. One Size Fits None

Article #3 – Cat and Mouse

Who isn't familiar with the iconic cartoon characters of Tom and Jerry? The cat, Tom, tirelessly defends his owner’s house, while Jerry, the clever mouse, seeks out tasty treats from the fridge. And more often than not, Jerry comes out on top. But what's the connection to immigration reform or transformation? We'll dive into that shortly.?

When we look at immigration reform, as I discussed in my previous article, it's clear that migration volumes worldwide are on the rise. With increasing volumes, governments are eager to attract individuals with the right skills and talents, whether it's to construct buildings or provide care in hospitals or knowledge workers. It's essentially a battle for talent, as countries compete to entice the same highly educated and skilled individuals to their shores. This fierce competition underscores the importance of attracting the right skills to foster growth and prosperity in each nation.

Speed of decision making as a competitive advantage

Making the decision to relocate to another part of the world is a significant one, and nobody wants to endure lengthy waiting periods to know whether their application will be approved. This is where the competition among countries for talent becomes evident, with speed of getting a decision being one of the primary criteria for visa applicants to target country A or country B. And while the visa application process may seem straightforward at first glance – applicants provide information, government officials assess it, and a decision is made – there are various factors that contribute to the often-prolonged decision-making process.

One key factor is the ability to verify the authenticity of the documents submitted by applicants. For instance, individuals applying for a worker visa typically need to submit a range of documents, including their university diploma or school transcripts, proof of good health to avoid burdening the host country's medical system, and evidence of sufficient funds to support themselves while in the country. Identity information is also crucial, as it facilitates risk assessments and ensures that individuals are not on any restricted lists. All these documents must be submitted to the visa processing agency, a process that traditionally involved mailing physical copies.

Some agencies have now transitioned to digital platforms, allowing applicants to submit scanned PDFs electronically. While this digitization streamlines the process and enables paperless workflows and work re-distribution, it does not fundamentally change the responsibility of the officer to check the veracity of these documents.

Opportunities with Artificial Intelligence (AI)

This is where the new technology of Artificial Intelligence comes in. And in my view, it can be a massive game changer for immigration organizations. Using artificial intelligence responsibly, we can have the system perform a first assessment of the veracity of certain documents.

The great thing about artificial intelligence is the step-up it provides from existing systems that interpret documents. A traditional system would scan a document using OCR (Optical Character Recognition) and would only be able to interpret if it has seen the exact documents before. For example it would know how to read a bank statement from bank X, but not bank Y if it has not been trained with this. Or picking up on our example of Tom and Jerry, it would only be able to tell that a picture is a cat if it has seen that particular picture of that particular cat before.

Artificial intelligence learns by itself using machine learning and lots of sample data. By feeding it lots of pictures of cats, it would learn what cats looks like, and would be able to determine with a high degree of certainty whether an animal is a cat in a brand new picture, even if it has never seen that particular cat before.

The same logic applies to interpreting documents for a visa processing application. Suppose a person provides a bank statement of a bank to provide they have enough funds in their account. Now there are thousands and thousands of banks in the world, so it would be hard to pre-program every single bank statement there possibly is. With AI it will be possible to massively shorten the duration of interpreting this. Because even though the system hasn’t seen this particular bank statement before, it has seen thousands of bank statements before and it will be able to assess whether the document is indeed a bank statement and whether the person has enough funds in their bank account.

Traditionally, immigration organisations (and governments in general) have responded cautiously to AI, and rightfully so. Nobody wants to run into the “Little Britain” situation whereby the “Computer says No”, without a clear rationale and traceability to regulations and legislation. Time and attention will need to be dedicated to train the AI to make sure there is ethical and transparent use without bias and completely removing the "human" touch on cases, as cases are people after all.

The next frontier: Generative AI

Most people will have heard about the next evolution of AI called Generative AI or GenAI. So what is the difference here? Where ‘normal AI’ can recognize patterns, GenAI can create them (hence the ‘generate’ aspect).

See this picture of a cat? It looks like a lovely furry kitten. The problem is, it does not actually exist in real life. It is completely created by GenAI. By drawing from previous instances, GenAI can conjure up seemingly realistic images, such as a lifelike depiction of a cat, even though the cat itself doesn't actually exist. It's akin to a digitally crafted illusion, crafted by the capabilities of gen AI. This is known as Text to Image capability of Gen AI. And as recently as 3 weeks ago, OpenAI launched their SORA product which provides a Text to Video capability enabling the creation of minute-long video based on a text prompt.

This remarkable technology also harbors a darker side as it could be exploited for fraudulent purposes. Individuals with malicious intent could use generative AI to fabricate convincing replicas of sensitive documents, such as bank statements, university diplomas, or health certifications. While these falsified documents may appear genuine at first glance, they are mere digital constructs devoid of authenticity. This ethical quandary underscores the dual nature of technological advancements – they can be harnessed for both beneficial and nefarious ends.

So, how do we address this dilemma of computers becoming adept at both interpreting and generating documents? On one front, existing technologies play a role in detecting forged documents by scrutinizing elements like the compression ratio in JPEG files for signs of tampering. However, such methods may prove less effective in combating the emergence of entirely new, digitally fabricated documents created by generative AI.

As gen AI produces fabricated images with consistent JPEG compression, traditional detection techniques may falter in distinguishing between genuine and counterfeit documents. Thus, as technology advances, so too must our strategies evolve to navigate the evolving landscape of digital deception.

Back to Source

Another innovative approach to identifying fraudulent documents is what we refer to as "back to source." Imagine if instead of merely interpreting a scanned document that might have been tampered with or is completely created by GenAI, you could go back to the source of the information on that document. For example, let’s say that for a university degree, immigration organizations had direct access to a university's database via an API. This would enable officials to verify an individual's graduation directly from the source system, eliminating any doubts regarding their educational credentials.

Similarly, rather than relying on manual passport verification, accessing a country's passport database directly could streamline the identity verification process for visa applications.

However, implementing such measures will pose significant challenges. For instance, granting foreign governments access to sensitive databases may raise privacy concerns, and establishing connections with numerous banks and educational institutions globally would require substantial effort.

As a result, we may find ourselves relying on a combination of strategies for the foreseeable future. Nevertheless, it's undeniable that these technological advancements are here to stay. By leveraging such innovations, immigration organizations can significantly expedite visa processing, thereby enhancing a country's ability to attract qualified applicants and foster societal growth and prosperity.

Cat and Mouse – the saga continues

Returning to the analogy of Tom and Jerry, we can draw parallels between the perpetual chase in the cartoon and the ongoing battle between computer systems in the realm of artificial intelligence and generative AI. Just as Tom and Jerry employ cunning strategies to outsmart each other, computer systems are utilized both by authorities to thwart fraudulent activities and by malicious actors seeking to exploit vulnerabilities. This cat-and-mouse dynamic is expected to persist for the foreseeable future.

For immigration transformations, this means that immigration organisations will need to keep evolving. There will need to be a constant assessment of the evolution of opportunities and threats and the swift introduction of relevant new technologies to stay ahead in this ever-evolving technological landscape. Above all, transformation will become an ongoing activity so the immigration cat stays ahead of the malicious mouse.

Link to Article 1: Agility is Key

Link to Article 2: One Size Fits None