Imagine a prospective skilled migrant from SEA submitting a visa application on a Sunday evening. Within hours, an AI-powered system has verified documents, assessed basic eligibility against risk profiles, and generated a preliminary approval recommendation. What once took days or weeks of manual review now moves at digital speed. A result could be out as soon as Monday morning.

Is this too far-fetched?

For Immigration New Zealand (INZ) officers, this is no longer science fiction. While still in development, it is the unfolding reality of platforms like ADEPT and other products in development.

Pair this with the recent announcement that the New Zealand government aims to pursue ambitious savings of $2.4 billion over four years through the reduction of around 8,700 public service roles. The magic bullet? AI. AI is positioned as a central enabler. From the information available, the impact exists on Immigration New Zealand as such plans point toward automating up to half of visa roles (around 500 positions) by 2032, shifting routine work away from humans.

Hopefully, this trend is not about replacing officers wholesale but about augmenting their capabilities. Presumably the savings are where AI can handle the repetitive, data-intensive tasks that currently consume significant time, freeing skilled professionals to exercise judgment where it matters most.

Yet, these efficiencies come with caveats. Chief among them is the urgent need to train and prepare officers to collaborate effectively with AI systems. Without this investment in human capital, the promise of transformation could falter.

I like to think of myself as an early adaptor of generative chat. I remember when I prepared a seminar on international child migration back in March 2023. A central point of the seminar was explaining the dangers of advice given by AI, generating what on the surface appeared correct but was functionally wrong legal advice. The message was to beware clients seeking their own research before seeking professional advice. Nothing new; it is similar to when doctors have patients who google their own symptoms.

What struck out to me was that one of the comments from the delegates stated:

• Speaking from experience [sic], very practical. Good topic, a bit too short and the 'Chat GBT' [sic] stuff was confusing (in terms of whether it is correct).

Looking back now the change is hard to believe. 3 years ago a presumably well-informed and intelligent immigration professional was confusing Chat GPT with Chat GBT. Now? It will be hard placed to find someone who works in the knowledge industry who does not know what it is.

Value of AI can be found in 'easy work'

By now I think regardless of where you stand on the spectrum towards AI we have to concede that on some level, AI is valuable. AI excels at pattern recognition. It can do data matching very quickly, noticing non-standard points. If everything is as it should be, then it can really be the future of standardized decision-making. It’s precisely the elements dominating much of routine visa processing. Under the Our Future Services programme found at https://www.immigration.govt.nz/about-us/our-strategies-and-programmes/our-future-services/  INZ has advised that it is consolidating multiple legacy systems into a single cloud-based ADEPT platform. Touted as the next big thing in the Immigration space for New Zealand, apparently this will enable automated document verification, risk flagging via connected analytics, initial triage of low-risk applications (such as many visitor or student visas), and even end-to-end automated approvals for straightforward cases.

With such improvements in productivity, it is easy to see how the seductive allure of outsource more and ever more to the mystic powers of technocratic improvement. And yes, AI has made an impact on workflow which cannot be denied.

An ideal balance?

There are many components behind a decision being made. In easy situations that do not require judgment, then yes. But what about in complicated situations which may require a humanitarian approach? The reality is that life is messy. Things change and develop in ways which are not expected. In the real world (and not a computer simulation) things can happen which appear implausible. I have certainly seen cases which appear fanciful and would likely be dismissed by a computer program as false.

An ideal outcome is one where an officer can be trained to use AI tools as a tool (and withstand the temptation to hand over their discretion to the algorithm). Enhancing decision making by using AI-tool as a form of spotting for when a risk flag deserves deeper review. The role of the officer can then instead be developed in understanding nuances AI might miss, and maintaining empathy in communications.

The risk though still remains. As Microsoft has discovered, efficiency savings in AI tools come at a cost in a myriad of other hidden factors. My view is that the required training to keep the balance of AI and human-ness is too high, at least for now. As with any role requiring use of judgement, it is not a one and done workshop. Something of such lofty aspirations is going to need continuous investment in professional development, simulation exercises, and guided by clear governance frameworks. In the short term, aiming for this development will likely cost more than simply hiring a person to do things the old way. With the budget available on the human resource side of things, it appears an overconfident expectation.

So where does that leave us? For now though the decision power remains or at least appear to remain human made to preserve accountability. Immigration decisions have the power to drastically alter a person (or family)’s live. Fundamentally something feels off about a machine making decisions about a human’s fate and future.

Cynically, the human accountability is likely to remain in the future. After all, someone has to take the blame when things go wrong, and you can’t fire a computer program.

Disclaimer:

This article is general commentary only and is not legal advice. Always seek tailored legal advice before making immigration decisions.