Get all your news in one place.
100’s of premium titles.
One app.
Start reading
The Guardian - AU
The Guardian - AU
Technology
Sarah Ayoub

Recruitment by robot: how AI is changing the way Australians get jobs

Robot selecting job candidate
The elements of the hiring process that were once the domain of human beings – resume screening and preliminary interviews – are being outsourced to AI. Photograph: AndreyPopov/Getty Images/iStockphoto

When Anisa* graduated from her second degree, she felt “fairly confident” that her postgraduate studies, double-major undergraduate degree and years spent balancing two volunteering roles and a part-time job would account for something in the job market.

She applied to every entry-level or junior role she could find in her industry, tracking each application’s outcome on a spreadsheet. Before she knew it, “applying for jobs became a full-time job”: in total, she applied for, and was rejected from, 350 jobs before finally landing one 18 months later. And she believes that AI – in particular its use in screening applications – is a huge part of the reason.

“The rise of third party, AI-run digital online forms are a huge pain point for so many jobseekers,” Anisa says as she recalls uploading CVs alongside filling in digital forms.

“I can count on one hand how many personalised interview processes I had where there was a face I could put to it.”

Although the uptake of AI in the recruitment space in Australia is fairly new, it is not uncommon. The elements of the hiring process that were once the domain of human beings – resume screening and preliminary interviews – are increasingly being outsourced to AI, with researchers estimating that there are now more than 250 commercial AI recruitment tools being used in Australia. As of last year, one in three Australian organisations reported that they had utilised AI tools while filling positions, though some researchers have deemed this a “high-risk” activity.

John Shields, professor of human resource management and organisational studies at the University of Sydney Business School, says increasing use of AI in the recruitment sector is not necessarily a bad thing.

He believes that it’s all about “quickly and effectively managing huge bodies of information on candidates” and achieving “a strong alignment between individual and organisation, or individual and job”, mitigating the issues and cost of a mis-hire.

According to Sue Williamson, associate professor of human resources management at UNSW, employers are turning to AI because the typical, traditional recruiting process – application, interview, work test, referee contact – is “outdated, onerous and time-consuming” and does not always result in the best person for a job.

Jobseekers are increasingly being interviewed on video by AI.
Jobseekers are increasingly being interviewed on video by AI. Photograph: Westend61/Getty Images

“Managers are increasingly time-poor and looking for ways to shorten the recruitment process,” she says. “They may be turning to AI to help streamline the administrative processes or to assist in vetting potential candidates, for example, by ‘assessing’ resumes or providing candidates with work tests. Video resumes are also becoming increasingly popular and these can also be assessed by AI.”

But it’s a sore point for many jobseekers, raising issues around fairness and merit. Recent research has revealed that AI-assessed job applications reinforce biases against women and cultural minorities. In Australia, the Merit Protection Commissioner issued new guidance for employers in the public sector after overturning 11 promotional decisions made by Services Australia in the 2021-22 financial year because they were made using AI-assisted and automated selection techniques that involved no human review, leading to “meritorious applicants missing out on promotions”.

“A lot of research has been conducted about biases in recruitment and selection, with one of the most well-known being that of Amazon, where programmers loaded up the AI [resume screening tool] with resumes from men,” Williamson says. “The AI then preferenced future resumes from men, disadvantaging women. Studies have also shown that candidates with non-western names, including Arabic and Asian names, also receive far fewer invitations for interview than candidates with western names.”

‘Job search depression’

Amplifying candidate concern is the lack of transparency around the recruitment process – or specifically, how they will be reviewed. Williamson says that Australia does not have specific laws that require candidates be informed about AI screening processes.

“This is potentially problematic as it may lead to indirect discrimination against groups of jobseekers,” she says. “Requiring candidates to undertake online tasks as part of recruitment [for example] may disadvantage those with limited manual dexterity.”

Anisa cites the time spent on various tasks, the lack of feedback to applications and the emotional and physical toll of such a repetitive process, which often culminated in what she calls “brutal” rejections, as a source of frustration and resentment.

“[The process] was stressful, disheartening and draining in the searching and the applying, the numerous interview stages and [consecutive] rejections and then the hundreds of times I was simply ghosted,” she says.

With the average job search taking around five months, that resentment can also lead to “job search depression” which experts warn can manifest in increased procrastination, maladaptive coping (such as increased drinking, sleeping or disordered eating) and prolonged anxiety.

Lisa*, who has been in her industry for 15 years, applied for the same job at a different employer and found herself being interviewed on video by AI. She would be asked a question, given 60 seconds to respond and then recorded as she answered it. Frank*, a data analyst, remembers how “nerve-wracking” it was having to answer a 100-question “psych-test” before taking up his current role. It’s since turned him off any tasks employers expect of candidates at the first instance.

“I think the first point of call should always be a face-to-face interview to get to know a candidate.”

Anisa agrees.

“A lot of screening interviews are also done by timed ‘tests’,” she says. “One [that] I did for a global tech company involved written and … video responses. It felt really impersonal and I didn’t meet or speak to anyone – even through a phone call – before I got a generic rejection email. For another role we were given problem-solving maths questions, working out [the] cost of services after inflation, and a final open question about how much money you’d need to live the rest of your life comfortably without working.”

‘I think the first point of call should always be a face-to-face interview to get to know a candidate,’ says jobseeker Frank.
‘I think the first point of call should always be a face-to-face interview to get to know a candidate,’ says jobseeker Frank. Photograph: Aleksandr Davydov/Alamy

‘Biases and unfairness’

But Shields says that cognitive abilities tests are nothing new.

“No [candidate] can succeed unless they are quite literate or quite numerate so we have to be careful not to say that these things are the phenomenon of the age of the robot,” he says.

“What is new is the use of candidate-generated online videos where the responses are then analysed by AI to profile candidates and to rank them in terms of qualities or attributes that the AI platform is analogically programmed to identify as desirable or undesirable. Candidate-recorded interview is going to be an increasingly important part of a filtered process before a candidate gets close to a face-to-face or panel interview.”

Anisa has described the rejections during her job search as brutal; Lisa* says the situation is insane because the process should be two-sided.

“Initial interviews are vital in understanding more about the business and how you see yourself working there,” she says. “It feels egotistical of [a] business to think that a interview should be one-sided.”

But Shields says the technology can be effective for both employees (in aligning them for the right job) and employers (getting the most suitable candidate) if it is used properly.

“If the programming that determines the recommendations of the AI platform is itself biased, it’s garbage in, garbage out,” he explains. “There’s got to be close attention paid to [ensure] analogic rules for AI platforms are literally written to iron out biases and unfairness.”

So what’s a jobseeker to do?

Stick to employer expectations and be clear about their strengths when applying for jobs, Shields says, recommending resumes are concise, precise and accurate, “and more importantly the covering letter needs to be forensically focused on the selection criteria”.

He also warns jobseekers to be careful of their online footprint. “If you’re serious about getting that job, do not have anything in your media profile that might raise questions about your integrity, behaviour, your political views,” he says.

*Name has been changed

Sign up to read this article
Read news from 100’s of titles, curated specifically for you.
Already a member? Sign in here
Related Stories
Top stories on inkl right now
Our Picks
Fourteen days free
Download the app
One app. One membership.
100+ trusted global sources.