Experts are sounding the alarm about what happens when recruitment is increasingly controlled by opaque AI systems few people understand. There will be a price to pay, on both a societal and individual level. Hire a resume writer who knows about AI syntax. This story was in the Sydney Morning Herald, written by Stephanie Wood.
AI tools are becoming increasingly common in the hiring process – for both employers and jobseekers. Their benefits are being widely debated.
You’re fast asleep, snoring, dreaming happily that you’ve quit work, you’re somewhere warm, siestas, great books, cocktails at sunset. Suddenly though, something jolts you awake. Good morning, reality! Now you remember. Reality is that you’re unemployed and your redundancy dollars are going down while your living costs go up. Buying groceries brings on anxiety attacks. Ahead again today: drudgery, scanning jobs ads, LinkedIn to Seek and back again, writing new cover letters, tweaking your résumé, sending out a fresh round of applications to add to the dozens you’ve submitted over months.
But then you remember something else: your personal AI job-hunting agent – I’ll call “her” Mavis – has been busy since before dawn. Think of Mavis as a PA: While you were dreaming of sunset cocktails, Mavis scraped job boards for newly posted career opportunities deeply aligned with your unique experience, skills and proven track record, dug into your digital files to pull up your résumé and cover-letter template, then tailored them for each fresh application. Relax, go out and grab a coffee (if you can afford to); when you’re ready, Mavis’s work on your behalf will be waiting in your inbox, ready for you to check over and send off.
Don’t take for granted, though, that a human being will be sitting in an office somewhere waiting to read your application. Mavis, meet Bob.
Bob is going to take care of things at the employer’s end. Bob, as I’m going to call “him”, is one of a growing number of AI recruitment tools.
Bob and its peers don’t actually read resumes. Nevertheless, they can make decisions about job applications and act upon those decisions on behalf of hiring managers and recruiters.
These tools, variously described as chatbots, AI agents or AI co-pilots can, among their many superpowers, shortlist candidates and advance them to the next step in the process – or, simply, click, bingo, eliminate them. It’s not cold, it’s not brutal, it’s just business.
Applicant tracking systems (ATS), which collect and sort job applications and track candidates through the stages of a hiring process, have been around for more than a decade, largely in a clunky supporting role to human HR staff. Increasingly, however, employers are adopting more sophisticated AI-powered recruitment tools, which take on tasks once performed by human recruiters.
These more “intelligent” tools can drive and connect almost every step in the hiring process, from researching, analysing and shortlisting candidates and identifying potential applicants beyond those who have applied directly, to conducting preliminary screening “robo-interviews” and psychometric tests, carrying out background and reference checks, answering applicants’ questions and scheduling interviews.
About 80 per cent of Australian companies’ report using AI technology at moderate or higher levels in recruitment, according to the 2025 Australian Responsible AI Index.
In the opposite corner, fighting fire with fire, many job-hunters now use generative AI systems such as ChatGPT, Claude and Gemini to draft and/or cross-check job applications. But in tacit recognition that the hiring process has become too machine-mediated for ordinary humans to negotiate unaided, the savviest of jobseekers in the tech sphere, are pioneering new ways to navigate the employment market. They’re using what’s called “agentic AI” to build personal AI job-hunting agents like Mavis to automate multiple elements of the dismal process.
I chat with a friend, a senior HR leader, and tell her that I’m working on a story about job-hunting in the age of AI. She sighs. “Recruitment will become the bots just botting each other soon, little robots yelling at each other,” she says.
Older jobseekers face issues including ageism, income security, under-employment and a soul-eroding sense of irrelevance. Ever more sophisticated AI-powered recruitment systems are contributing to the challenges and stress all job-hunters experience, not just older ones. As a New York Magazine noted in an article earlier this year, “The job search has become a humiliation ritual.”
Recruiters widely acknowledge that the systems can make poor decisions: algorithms can determine an excellent candidate is not a sufficiently close match for a role and eliminate them before he or she has a chance to get near a human.
Anna Whitlam, CEO of WhitlamCo executive search and advisory firm in Melbourne, offers the example of a senior corporate affairs leader, “a no-brainer candidate”, who applied for a job but was immediately rejected by an automated system. “Time and time again what you will hear is a senior candidate not even getting an interview,” Whitlam says.
Amanda Bowler realised when she was 40 that her arms weren’t going to let her keep working as a hairdresser until she retired and, in 2018, she started a psychology degree. “I really love people and the mind and behaviour and how it all works,” Bowler says via Zoom from the granny flat behind a friend’s house in the central Victorian town of Trentham, where she’s living temporarily.
Bowler finished her honours year in psychology last year. Before embarking on the postgraduate study required for clinical practice and tested the job market in areas including aged care, research and program delivery, hospital administration, disability support and child protection casework. She estimates that over the past six months she has submitted 50 job applications. “Basically, crickets. I have not had one face-to-face interview.”
In such a dehumanising job market, there will be casualties: the technologically unsophisticated, the anxious, the sensitive or the timid struggle. Older workers who pre-date the LinkedIn-buzzword-optimised, keyword-dense professional identity frequently find themselves in unwanted early retirement or the driver’s seat of an Uber.
AI systems are not generally friendly towards people who have had varied career and life experiences and built a wide array of skills rather than a specific, highly honed few that can be encapsulated in corporate-speak, or those like Bowler, who might be trying to launch into a new industry.
Think of it as a giant game of word-matching: AI recruitment systems scan at a granular level for keyword and terminology matches between a position description and an applicant’s experience. If, for example, a job ad for an aged-care worker asks for someone with experience in “case management”, “holistic health plans” and “person-centred care”, and an applicant like Bowler describes broadly similar experience using different language, a human recruiter reading her application might recognise her aptitude for the role.
But the AI system has no imagination, no interest in nuance and debatable judgment. Bowler’s application is likely to be screened out because she has not used precise keywords and terminology.
Recruiters can also instruct the systems to prioritise candidates who have one particular credential or experience in a particular niche or sector and downgrade those who don’t. For example, a car company looking for a CFO could tell an AI recruitment system that it is only interested in people who have worked for Toyota and Subaru; someone who might match every other requirement for the role but who has only worked at Hyundai won’t get a look in. Such is the opacity of the AI systems that applicants will never know why. Transparency has always been limited in the job application process but at least in the past, decisions were made by humans and potentially explainable.
Others face ageism and the searing issue of finding relevance and meaning in life when there’s no job to provide it. Cameron McDonald, 63, has not been able to find work since his position as a senior business development specialist with a Canberra consultancy was made redundant in 2022. He says he wasn’t actively chasing work, but when he was approached via his LinkedIn profile to put his hat in the ring for a handful of roles, he submitted applications. He got through to the interview stage for two of the positions but was eventually eliminated. McDonald blames ageism. “The people who got the job, I could see, were 10 to 15 years younger than me,” he says. “It didn’t make me feel too good; the most upsetting thing was the fact that I knew that I had a hell of a lot to offer.”
Although financially secure enough not to work, McDonald thought a little pocket money might be handy and, in late 2025, he applied for a service position at Bunnings. He reached the second stage – a recorded one-way video interview, often called a robo-interview. These increasingly common, torturous sessions require candidates to talk to their blank screen recording video responses to a series of questions, with the footage later reviewed by recruiters, AI software or both. One jobseeker tells me that he felt he always looked flustered during the interviews as his eyes were forced to dart between the question and his computer’s camera.
“I do not agree with them,” says Sydney recruiter Stephen Coldicutt. “I don’t think it’s great for the candidate at all, they don’t allow people to be natural.”
His issues might or might not have had anything to do with the AI hiring process, but experts worry that automated systems make it harder to detect and challenge existing forms of discrimination. The opacity of the technology is a massive problem.
There are serious risks of discrimination when employers use AI in hiring. You could put anything into that back end: you could put in that you don’t want a woman, or you don’t want anyone over a certain age.
The Stanford Digital Economy Lab have found evidence that when large numbers of employers use the same AI hiring systems, the same algorithms can produce repeated outcomes and biases across the labour market, potentially disadvantaging the same kinds of candidates across multiple employers.
AI is a double-edged sword. Recruiters face a new challenge: determining the authenticity of applicants, Nardone says. “It’s harder to distinguish because our candidates are utilising AI themselves for their own applications. AI is still not the perfect tool … it still requires the human element.”
But AI specialist Abhijay Arora Vuyyuru does not believe the future of job-hunting lies in ever greater torrents of applications written by Claude or ChatGPT washing into recruiters’ inboxes. Instead, it’s something else altogether.
Enter Mavis, your personal AI job-hunting agent.
“The shift we are seeing is from just a simple prompt that you’d give to ChatGPT to something more complicated where you’re creating an automation that runs at a given time,” says Vuyyuru, who works on AI products at Alphabet-owned YouTube.
He became interested in how a job search could be assisted with agentic AI – systems that can plan, act and use tools with some autonomy – after graduating in 2022 from Harvard Business School. Frustrated by the time he spent applying for internships, he built a personal job-hunting agent which can work across a range of job-hunting tasks, from scraping job boards to tailoring résumés.
He believes that, within the next few years, those looking for work who don’t have a personal AI agent will be at a disadvantage because people who do “will be so much more proficient in applying for jobs”. But Vuyyuru’s predictions don’t stop there. “AI agents are going to be everywhere; it’ll be an interesting space.” He says communication between employers’ AI agents and job-seekers’ AI agents will be “big”.
An absurd future is upon us. Think of it as a case of “I’ll get my AI agent to talk to your AI agent.”
Matt comes on to my computer screen with a broad smile on his face. I tell him that he looks like he just got a call to say he got the job (he didn’t). “Every day is a wonderful opportunity to meet or exceed customer expectations while delivering outstanding value,” he says. I do a double take, then realise he’s poking fun at management speak.
Until late last year when Brisbane-based Matt, 51, stepped away from work to devote time to his neurodivergent daughter, he had spent nearly two decades in executive positions as an operations expert across a range of companies. He embarked on the process of returning to the workforce about three months ago. The experience, he says, has been “horrible”.
Matt’s big smile fades. He has applied for about 200 jobs and, in some cases, had AI rejections to applications within 30 seconds of submitting them; he’s had to provide his date of birth and refused to do robo-interviews and been ghosted by recruiters.
He has developed some strong opinions about younger members of the recruitment industry. “Their understanding is really single- or two-dimensional at best,” he says. “They go, ‘Oh, we spoke to the client and the client said they want XYZ, do you have XYZ on your CV?’ If you don’t have XYZ and you’re not matchy-matchy, you’re not the right person.”
His comment is a grave reminder of the humanity at the heart of job-seeking. Life stressors run alongside the process, emotions run high, important life decisions go on hold, things go nowhere.
The creeping encroachment of increasingly powerful AI recruitment systems adds a new layer of agony. All this brisk, automated efficiency, hundreds or thousands of human hopefuls sorted, filed, rejected, processed. …
Ironically, all the job-hunters I’ve interviewed who are dealing with AI recruitment systems say that relationships, human contact, lie at the heart of their work – why they love it, why they’re good at it. “The way I operate, relationships are at the core,” Matt tells me. Amanda Bowler, who as a hairdresser spent her days listening to her clients’ stories with kindness and empathy (who hasn’t poured their heart out to their hairdresser?); Phoebe, whose corporate affairs work relies on her ability to form relationships and collaborate with stakeholders to solve problems or advance opportunities.
If there is a small shaft of light, it’s that the experts I speak to tell me that the more AI systems are used, the more humans will be needed. For a start, recruiters and career coaches agree that intense networking with other human beings, the a-coffee-date-a-week approach, is still the best way to get a job.
“Self-advocacy and personalising your approach to a job search is actually quite a game-changer and takes you out of being part of the mass application pool,” says Jannine Fraser, group CEO of Melbourne-headquartered career transition and coaching specialist Directioneering. Fraser says that if a job-hunter has been out genuinely connecting with people – industry contacts, potential employers, recruiters – they might already have met someone in some position of influence before any job hits the market. “You’ve already had a conversation.”
Conversely, as AI makes it easier for people to misrepresent themselves in their job hunt, humans in positions of authority will be needed to identify the authentic candidates. “This may sound a little bit counter-intuitive, but human relationships will become more and more important,” Abhijay Arora Vuyyuru says. “I would think that a lot of hiring is going to happen in person; a lot of these AI systems are getting so good that [an employer] would want to test whether a person is actually able to think for themselves or has just leveraged AI for their application.”