The idea that your next job interview might be conducted by artificial intelligence would have sounded far-fetched five years ago.
Today, it is a reality, and it is part of a much wider shift of automation reshaping recruitment.
Recent reporting suggests that a recruitment software provider conducted more than 1,000 AI-led interviews with New Zealand job seekers in a single month.
Before candidates make it to these interviews, employers are using AI to scan, rank and shortlist hundreds of CVs at speed.
The result is that a candidate may have their CV screened and their first interview completed before a human has meaningfully engaged with their application.
For businesses, the appeal of automation is obvious. High-volume screening, faster hiring timelines, and the sense of a more consistent process.
But the risk is equally obvious if AI is treated as a substitute for careful human judgement.
And when it goes wrong, the legal and reputational consequences still sit with the employer.
Reliance on these tools is the result of hiring managers and HR teams experiencing real pressure not only to keep up with but also to make the right decisions about hundreds of potential candidates for limited roles.
AI tools, which promise consistency, speed and therefore cost savings, may appear attractive as the remedy to these challenges.
But the speed at which AI has moved from the margins of recruitment into the mainstream may have outpaced both public understanding and scrutiny.
A central selling point of AI tools, especially as they relate to recruitment, is that they remove human subjectivity.
Algorithms can apply the same criteria to every CV, interview answer, and recorded response.
Yet consistency does not automatically result in a fair, or the right, outcome.
AI systems learn from the data upon which they are trained. This ‘training” is based on human reasoning, which may perpetuate harmful and toxic biases.
As it relates to recruitment, AI tools are typically trained on role criteria and an organisation’s historic hiring data.
If past patterns tended to favour certain schools, career paths, and personalities, those preferences can be replicated at scale.
What appears neutral on its face may systemically disadvantage candidates who do not fit the data profile the system has learned to reward.
If an AI screening or interview tool filters candidates (directly or indirectly) on characteristics such as sex, age, race, disability or family status, employers may unintentionally expose themselves to claims of unlawful discrimination.
As a result, internationally, AI recruitment tools are increasingly being treated as “high risk”.
The European Union’s AI Act, for example, sets expectations around transparency, human oversight and ongoing monitoring, signalling the direction of travel even for employers operating in other jurisdictions.
In New Zealand, if something goes wrong, the fact that the decision was made or influenced by a machine is unlikely to shift responsibility from the employer.
Additionally, job seekers report being unaware that early-stage interviews have been conducted by AI and uncertain about how their responses have been assessed, stored, or reused.
Recruitment processes require gathering a large amount of personal information, triggering a raft of legal obligations under the Privacy Act.
Particular concerns arise if the tools analyse voice, language, behavioural cues, or other biometric data as they try to improve the product.
There is also a broader societal dimension that sits beyond strict legal compliance.
Recruitment is often a candidate's first real interaction with a prospective employer.
Being assessed by an automated system, particularly without prior warning, can feel impersonal, alienating and dismissive.
For some candidates, particularly those already struggling in the labour market, this may reinforce a sense that the process is stacked against them in ways they cannot change
In response, candidates are increasingly using AI to draft CVs and cover letters, tailoring applications to keyword screening and applicant tracking systems in an attempt to game the system.
Ironically, the more recruitment becomes automated, the more it risks incentivising candidates to automate their application, making meaningful human assessment harder.
However, none of this is to suggest that AI has no place in recruitment.
There is no doubt that, when used properly, these tools can support human decision-making and reduce administrative burden.
But, as New Zealand employers continue to experiment with AI-led hiring, they should be asking the right questions, including where this tool supports judgement and where it is replacing it.
Originally published by The Post
