Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode 138 - Are we trusting AI too much in hiring?
0 Playsin 14 hours
Episode 138 - Has the updates from RCSA board and council announcements. Early entry jobs and digital transformation in the government both in trouble. As well as results from the Seek SARA awards and an accolade H People received that very few other agencies have. Our question this week comes from a University of Washington study that highlights the issue when AI is involved in hiring. Are we trusting AI too much in the hiring process?
Recommended
Transcript
Sponsorship by Wingman Recruitment
00:00:10
Speaker
This episode of Recruitment News Australia is brought to you by Wingman Recruitment. the offshoring partner built by recruiters for recruiters. Created by H-People and the Wingman Group, their model gives you back your time by shifting administration compliance and sourcing to remote professionals so your consultants can focus on what matters, building relationships and making placements.
00:00:34
Speaker
Learn more at wingmangroup.com.au via the services tab.
RCSA Board Updates and Female Leadership
00:00:39
Speaker
This is the news for the November, 2025. I'm Clennett.
00:00:44
Speaker
And I'm Adele Last. We kick off news this week with the announcement of the RCSA Board of Directors update. We've got five returning board members to the RCSA board. They are Catherine Amani from OnQ Recruitment, Matthew Ustini from Technical Resources, Natasha Olson-Sito from OnTalent, Yelena Giro from Altera and Jason Elias from Elias Recruitment.
00:01:11
Speaker
And also breaking news, Adele Penny O'Reilly has stepped down halfway through her two-year term and Natasha Olson-Sito has been elected as the new RCSA president. Congratulations, Tash. And if I'm not mistaken, that makes four women presidents in a row, which is great news for the RCSA. And I'm sure Tash will do a great job.
00:01:36
Speaker
Further updates from the ah RCSA also include new appointments to their regional councils. For the Amrans and ANRA councils, which are the medical and nursing councils, there's been new elections to the Amrans council in Paul Darby from Wavelength and Phoebe Bardsley from Go Locum. And to the ANRA Council, Paul Chizik from Alliance and Michael Fernandez from Blueprint Nursing. New South Wales ACT Council. We have four people re-elected and four people new to council. Rob Esposito, Macrile Partners, Paris Matten, Healthy People, Rhonda Adams, Express Employment Professionals and Olivia Evans from People Fusion as the next-gen representative.
00:02:21
Speaker
And in Queensland, Northern Territory, new council member Cassandra Manthe from Lime Recruitment. SA Council, three people re-elected and new to council, Samantha Eddits from Randstad. In the Vic Taz Council, my own council, where I was re-elected. Thank you very much. We also have four new council members, Sarah Swiderski from Persol, John Borges from Spencer Ogden, Amanda Chisholm from Kaizen Recruitment and Blake Clark from Programmed as the NextGen representative.
00:02:50
Speaker
WA Council, one member returning and new to Council, Bronwyn Butcher BB Recruitment and Consulting, John Shattuck, Sudstrom Recruitment and Thomas O'Connor, HSS Recruitment as the next-gen representative.
00:03:04
Speaker
And across the ditch in New Zealand, two re-elected Council members and two new ones in Ian Scott from the Scott Collective and James Pomeroy from Pomeroy Group.
Challenges in Entry-Level Job Market
00:03:16
Speaker
A new snapshot from Anglicare Australia, Adele, shows that entry-level job opportunities are shrinking quickly and the competition for those jobs that are left is the toughest it's been in a decade.
00:03:30
Speaker
Yeah, the numbers are pretty confronting, Ross. Right now, for every single vacancy that requires minimal formal training or experience, there are 39 job seekers competing for that one role.
00:03:41
Speaker
Last year, it was 35 to 1, so the situation has deteriorated quickly. And out of those 39 people, Anglicare tell us 25 face significant barriers to employment, things like being long-term unemployed, having disability or limited English skills.
00:04:00
Speaker
Anglicare Australia's Executive Director Casey Chambers, not the singer, said this is the worst result they've seen in 10 years of producing this annual report. She also pointed out that many of these job seekers are now competing with people who don't face those barriers but are applying for low experience jobs just to manage the rising cost of living.
00:04:21
Speaker
So the competition is even fiercer than the data suggests. Well, federal government data shows job ads have been declining for three years with the steepest drop, 37%, in roles that traditionally have acted as a stepping stone into the workforce.
00:04:37
Speaker
Inc., checkout operators and cashiers. Those kinds of positions are shrinking because of automation and self-service technology. Chambers also warned that entry-level jobs overall are under threat and that the system is failing the people who need labour market support the most.
00:04:53
Speaker
It's not only retail, JSA Australia interviews show that employers in areas such as engineering are hiring fewer entry-level engineers because routine tasks are being absorbed by ai tools.
00:05:07
Speaker
There's no broad collapse across all entry-level career roles, but the current trend is of concern. Exactly. We're seeing fewer trainee and graduate openings in tech, more competition and higher skill expectations for juniors.
00:05:22
Speaker
And the nature of entry-level work itself is changing. New starters are expected to handle more complex tasks right away, judgment earlier and be comfortable with AI-enabled systems.
00:05:34
Speaker
So it does sound to me like the picture is clear and not that encouraging. Fewer entry-level jobs, tougher competition for those jobs and rising pressure on job seekers.
00:05:47
Speaker
So my conclusion is, Adele, unless Australia acts by rebuilding real pathways into work, supporting those facing the greatest barriers and preparing for the impact of AI, we do risk losing an entire generation of future talent before they even get their start.
00:06:08
Speaker
On to award news, the SEEK SARA Awards were run in Australia and New Zealand recently, and we've got the announcement of the winners. For Small Recruitment Agency of the Year, in Australia we had Cura Services and in New Zealand Recruitment Studio.
00:06:23
Speaker
Medium Recruitment Agency of the Year, Australia, Fuse Recruitment and in New Zealand Talent Army. For the Large Recruitment Agency of the Year, a name that's very familiar in this category, People to People and in New Zealand Stellar Recruitment.
00:06:39
Speaker
Recruitment Process Outsourcing Agency of the Year. There was only an award for Australia and the winner was Solved by Talent. A new Recruitment Agency of the Year was Thomas Executive.
00:06:52
Speaker
Outstanding Progress in Diversity, Equity and Inclusion Australia. The winner was Harrier Talent Solutions and Cultivate Recruitment was the New Zealand winner. Excellence in candidate experience, one in Australia by Horner Recruitment and in New Zealand by Trade Assist.
00:07:10
Speaker
Recruitment Leader of the Year, Australia, David Jam, Vertical Scope Group, New Zealand, Brian Keegan, Sprout People. Recruitment Consultant of the Year was Shane O'Neill from Civitas Talent and in New Zealand, Lewis Howes from ah RWR Health Northern.
00:07:27
Speaker
And innovation of the year, this was purely a New Zealand category, winner, are AWF Limited. Many examples of great performance there. Congratulations to all the winners.
Digital Talent Shortfall and Solutions
00:07:39
Speaker
Adele, another new report, this time from the National Digital Transformation Agency. Findings are pretty stark. Says the federal government is staring down a digital talent shortfall of more than 8,000 people in the next five years. That's a big warning sign.
00:07:55
Speaker
The 2025 APS Digital Workforce Insights report says the Australian public service will actually need to double its digital workforce by 2030. Demand for new technologies is rising fast and about 20% of the current tech workforce is set to retire in the next half decade.
00:08:13
Speaker
Says the APS has around 8,000 ICT and digital employees and the median age of those employees sits between 40 and 44. So it's not a young workforce, which obviously only adds to the urgency.
00:08:27
Speaker
And the agency isn't just saying hire more people. They're saying the whole approach needs to shift. The APS has to rethink who it hires and how it hires. That means opening more diverse pathways into digital roles instead of relying so heavily on university degrees and graduate programs.
00:08:46
Speaker
Yeah, exactly. The report suggests building stronger partnerships with vocational providers and to have a greater acceptance of micro-credential programs. And this will widen the talent pool. And this approach just simply mirrors what um many in the private sector are doing right now.
00:09:03
Speaker
It's interesting that one of the biggest recommendations is to rethink where people are hired. The report says we need to break the Canberra bottleneck, it's called it, Currently, 60% of APS tech workers live in the ACT. That's just a tiny fraction of the whole talent pool, five less than 5%, in fact, of the entire Australian ICT workforce.
00:09:25
Speaker
Which means the APS is competing for talent in the smallest corner of the country instead of tapping into the national park market. Not exactly a recipe for filling an 8,000-person supply shortfall.
00:09:37
Speaker
So here's the reality. The APS is facing a big challenge, but also this could be a huge opportunity. By opening new pathways into digital careers, embracing remote talent, tapping into the whole national workforce, it's probably a great chance for Australia to build a stronger, more diverse and future-ready public service.
00:09:56
Speaker
In fact, I think if we act now, the talent gap could become a catalyst for genuine transformation and not just a barrier to it.
AI Bias in Hiring Systems
00:10:03
Speaker
In other news, Adele, a University of Washington study presented last month at a conference in Madrid on AI ethics and society provided a real wake-up call for hiring teams. Yeah, looking at this research, Ross, they tested 528 participants and how they were reacting to simulated AI recommendations while screening applicants for 16 different jobs.
00:10:27
Speaker
All candidates were equally qualified men, white, black, Hispanic, and Asian, but the AI recommendations were programmed with different levels of racial bias. And what's striking is that when people chose candidates on their own or with neutral AI help, they picked white and non-white applicants at equal rates.
00:10:46
Speaker
But the moment the AI showed moderate bias, participants followed it. And with severe bias, they still selected the AI preferred candidates about 90% of the time.
00:10:57
Speaker
So what you're saying, Ross, is that the people responsible for checking for bias were just following what the AI suggested. Yes. That's really worrying because around 80% of organisations using AI hiring tools say they don't reject candidates without human review.
00:11:13
Speaker
However, this example shows the process is far from perfect. Well, lead author Kira Wilson put it clearly, unless bias is obvious, people were perfectly willing to accept the AI's recommendations. And that means bias systems produce biased hiring decisions impacting real people.
00:11:36
Speaker
And to finish, fantastic news for our friends at H People, Jory and Sam. H People ranked number two on the Australian Financial Review's 2025 Fast 100 list. This list recognises fast growing and established companies and to be eligible for the list, a company must have started trading before July 1, 2020.
00:11:57
Speaker
have recorded a minimum revenue of $10,000 in the 2023 financial year and a minimum revenue of $5 million dollars in the 2025 financial year. Number two spot. What a great result, H people. Well done, Sam and Jory. Congratulations.
00:12:21
Speaker
Question of the week. Adele, are we trusting AI too much in hiring? I think I know where this one comes from, Ross. We just read news around a University of Washington study around the use of ai in hiring and the fact that the human that is supposed to be assessing the recommendation from AI was just blindly following the AI and they had built in some racial bias into that AI decision making. So this is a really worrying concern in our space.
00:12:53
Speaker
And frankly, it's fairly typical of how people get into trouble with AI. So firstly, a non-recruitment example, we in Australia have noticed that Deloitte, the big consulting firm Deloitte, got into trouble with the federal government. They produced a report for the federal government. They charged ah nearly half a million dollars for this report. But unfortunately for them, it was found out the report made claims based on research that had never happened and papers that had never been presented.
00:13:28
Speaker
And that, I think, is exactly the problem because whoever at Deloitte was responsible for reviewing that before it was submitted to the federal government knew enough to know that enough of what was written was accurate, but not all of it. And because it had assumed no doubt that what they were looking at that AI had produced that they knew was accurate was accurate that everything else was accurate as well and that wasn't true. AI had hallucinated and that was very embarrassing for Deloitte. And this has the potential for us in recruitment within the context of specifically at the moment I suppose the heavy element of where we're using a lot of AI is in the shortlisting process in the data matching process. I know A lot of the major CRMs in our space are building this into the platform where you basically put your job description into your system. It runs through your data, looks at your candidates, all of the interview notes and comments and resume and all of the data points that you've collected about a candidate and matches up candidates to your job.
00:14:38
Speaker
But the concern about this is are there biases in the data, in the job description, in what you're collecting about the person? And then the AI tool that's being used, has it been trained with biases implicating its decision-making as well?
00:14:53
Speaker
Well, for sure. I mean, a classic example would be, hypothetical, but I'm sure has happened, that a recruitment agency recruiter who may be very successful has actively excluded candidates with no Australian experience. And as a result, records for that recruiter's jobs would indicate people who have all got Australian experience and maybe all have Anglo names or all went to prestigious universities. So if the AI is using that as effectively a training ground,
00:15:25
Speaker
then anything subsequent to that that it's asked to do with respect to those types of jobs, then clearly it's going to be riddled with bias. And I think our concern about this is in relation to the law's legislation and even the branding element of this, as you mentioned, was very embarrassing for Deloitte because people will argue this to say, well, there's biases in humans, right? So we're we're so worried about the bias in AI, but what about the fact that we are all biased ourselves individually? So what's the big deal? You know, who cares about that? But I think the reality of this is around the fact that if you as a recruiter are being paid to source the best talent and you're using a tool that produces a result that maybe buys or may isn maybe isn't even accurate, ah you're being paid for that. And so, Now there's this level of responsibility for you to make sure that is accurate. If you had biases in that process, I guess no one's questioning that, but it's when something else is deciding it for you. This is where we're getting into this very murky waters, right?
00:16:31
Speaker
Well, yes, and the AI, or AI generally, is deemed to be some sort of solution for bias. To say, well, AI is not bias. So therefore, when AI ah goes in and generates a long list and a short list and recommendations, it's going to be agnostic. It's going to be objective.
00:16:54
Speaker
But clearly that's not true if it's um drawing on biased data. And because there's this assertion that because it's AI, it's objective, then that's even worse because just like humans, it's flawed and obviously not in all cases, but this is not the first study. There have been other studies where similar things have been um um proven that effectively humans too easily accept biased recommendations or inaccurate recommendations from Gen.ai, mainly in service of efficiency.
00:17:31
Speaker
Yeah, this is the thing. We're really putting high value on the speed that this tool and and the you know the admin that it cuts out, these tools provide us as opposed to the accuracy of it. where We're you know valuing efficiency over you know quantity and speed over over the quality. but If you think about our trust, think about the trust equation of this. You know, most people still probably fall very much into the category of a bit suspicious about AI. I'm a bit untrusting of this technology. I don't know if it's right.
00:18:05
Speaker
But we very quickly jump the fence, Ross, to the other side. We use it a few times. We we might run ah a search like this. We might verify it and check it once or twice or three times. But once we kind of go, oh, this feels about right, yeah, that's the kind of candidate I would produce it it would find for me, then we run with it blindly. We don't go back and repeatedly check that it's right.
00:18:27
Speaker
I mean, classic example, not quite the same as what you're talking about. But I think about two months ago, I asked ChatGPT, what does... um recruitment expert Ross Klenert say about, I can't remember what it was specifically, ah but it was obviously something to do with recruitment of the things that I've written.
00:18:47
Speaker
And it produced a response of which um I think four or five of the things were accurate and at least three of the things were completely inaccurate. ah Not things that I had said and certainly not things that I believe. So that sort of gave me a bit of a wake-up call in the sense that um AI can be mostly right, but it doesn't mean it's entirely accurate.
00:19:09
Speaker
So that's a bit the equivalent of, you know, you did ah a recent Google search, right? used to Google ourselves to see what was out there about us. You've asked ChatGPT to say, what would Ross say? And this is the interesting thing. Like you're saying, it's providing some correct data, but it's padding it with incorrect data or hallucinations or made information. Hallucinations,
00:19:30
Speaker
um That's the stuff. and And you've got to know where to pick pick out the differences, right? Precisely. Because if someone was, like ah if a journalist or blogger was writing an article about me and had put that into chat g GPT and got those results, then ah she or he could have just written all that. And if they weren't checking the sources, then they'd be attributing opinions that supposedly I have to me that were not true.
00:19:57
Speaker
And this is where I think a lot of people who are warning about the cautions of AI and saying, just be careful, ah you know, just don't trust it blindly is where we're seeing the results in these kinds of reports and research and and some of the examples we've just noted today. So it's still got a long way to go, right? We're such early days. And yes, it's moving really fast, all of the AI capability and the AI functionality, but you know, we we're really trusting it so early and we it's really dangerous.
00:20:28
Speaker
Well, I mean, one thing that I've heard more than once that I think is a very good mindset to have is treat whatever AI produces as a first draft.
00:20:40
Speaker
And from that, you do a lot more work because no one who's serious about quality work would produce a first draft and say, well, that's good enough for whatever it is, a first draft of an ad or a first draft of an essay or a blog, they're going to go back and do more work.
00:20:58
Speaker
And that's, I think, a very powerful mindset to take in terms of this needs two or three or four more revisions, which should include checking, statements that are made to ensure that there's source material. So it's substantial and it's accurate and it's drawn from facts, not from hallucinations.
00:21:18
Speaker
Yeah, look, I'm one that's, you know, a very early AI adopter. I love the technology. I'm using it every day. I'm researching tools all the time, working um with an organisation um with AI built in all of these things.
00:21:33
Speaker
But you still have to approach all of this with a really healthy healthy scepticism. And humility of um i don't know everything and I shouldn't assume that AI does as well because once you've damaged your reputation and clearly Deloitte certainly, I believe, have damaged their brand and reputation. So once it's been damaged, like there's a lot of work to recover that.
00:22:05
Speaker
And this ah efficiency dividend that AI can supposedly or is supposedly producing is like fool's gold. It's like, well, there's no point being faster to looking an idiot. It's like, you've got to be slower, take your time and ensure that you're not trashing your own reputation.
00:22:28
Speaker
So maybe we've got good news for recruiters then, Ross, that ah our jobs are safe for a while yet until we can really know that the AI is accurate, is free from bias, and is you know really genuinely using the data correctly, that we've still got a job, I suppose.
00:22:47
Speaker
And as long as our clients are making mistakes and we are not, or certainly not to the same degree when it comes to the use of AI in hiring, then I think we've still, we as a recruitment industry, still have a very bright future.