Latest / The Morning Edition / 'The CV is dead': How to get a job in the age of AI

Transcript

'The CV is dead': How to get a job in the age of AI · The Morning Edition

0:04
S1: If you've applied for a job in the last few years,
0:07
S1: you may have been humiliated by an AI program. Maybe
0:11
S1: while it interviewed you or throughout your CV. But lately,
0:14
S1: as AI recruitment programs have become way more sophisticated, so
0:19
S1: too have job hunters enlisting AI agents of their own.
0:24
S1: I'm Samantha Selinger Morris, and you're listening to Morning Edition.
0:27
S1: From The Age and The Sydney Morning Herald today, writer
0:32
S1: Stephanie Wood on new ways that AI can discriminate against
0:35
S1: workers and the best way to nab a job. It's
0:39
S1: September the 1st. Welcome, Stephanie, to the podcast.
0:46
S2: Hello. Thank you.
0:47
S1: Okay, let's just start off with the basic facts. Can
0:50
S1: you walk us through briefly, how does this AI recruitment
0:54
S1: system actually work? Like what are the main mechanisms or
0:57
S1: systems at play, I guess?
0:59
S3: Well, it's intensely complicated. It took me a long while
1:02
S3: to actually figure it out. The first thing I should
1:06
S3: say is that there have been AI recruitment systems for
1:09
S3: some years now. It's not something that's just suddenly new.
1:13
S3: What is changing is the sophistication of these models. They
1:18
S3: are taking over whole chunks of the process, whereas in
1:22
S3: the past they were largely sifting and doing a bit
1:26
S3: of filtering, but not really fitting into a recruiter or
1:30
S3: an HR person's entire process. So my understanding of it is,
1:36
S3: and everyone will tell you a slightly different way that
1:39
S3: they work is that, uh, whether it be an HR
1:43
S3: person within a company that's doing the hiring or whether
1:46
S3: it is a recruiter in a recruitment company, uh, they
1:49
S3: will have these very complex AI platforms with all sorts
1:53
S3: of names. One is called Winston from one company, an
1:57
S3: AI companion and hiring hero, is how their marketing material
2:01
S3: describes Winston, but they take a whole lot of steps
2:05
S3: in a process to recruit someone and take over, basically.
2:09
S3: So that is when you apply for a job, your
2:12
S3: application goes straight into the platform and often is filtered
2:17
S3: and sifted and keyword analyzed immediately. And I have spoken
2:21
S3: to people who's put an application in and within ten
2:25
S3: minutes they got a rejection. Because what that AI recruitment
2:29
S3: system is doing is it's scanning your cover letter and your,
2:33
S3: your CV for really micro keyword matches. So if, for example,
2:41
S3: an application calls for very specific set of skills, the
2:45
S3: idea is that in your application and cover letter, you
2:48
S3: need to use the exact same keywords as the job
2:52
S3: description is calling for. The position description is calling for,
2:56
S3: it's not going to be to your benefit if you
2:58
S3: put in similar skills or adjacent skills. The wording has
3:03
S3: to be just so, so precise because the AI system
3:07
S3: doesn't really have any concept of nuance, and it's looking
3:10
S3: for people who can exactly fit in boxes, basically. And
3:15
S3: if you don't, then it will eliminate you very quickly.
3:17
S3: But beyond that, these processes will possibly send out a
3:21
S3: response saying, thank you for your application. We're considering your
3:23
S3: application right now. It could set up interviews, this platform,
3:28
S3: and at this point, no human hand has even entered
3:31
S3: the process. A lot of people are talking about having
3:34
S3: experience now, those interviews that are early in the screening
3:37
S3: process where you're talking to a blank screen and you
3:39
S3: might see questions flash up on the screen, and you
3:42
S3: need to answer those questions to your blank screen. An
3:45
S3: incredibly discombobulating. And if you're already nervous, it's just such
3:49
S3: an unnatural interface, isn't it?
3:51
S1: Okay, now, you mentioned there that of course this could
3:53
S1: be discombobulating, which I think anyone listening can only imagine,
3:57
S1: you know, being interviewed not by a person, but of course,
4:00
S1: words flashing up on screen. But you've interviewed numerous people
4:04
S1: who've been through these processes, and they don't just describe
4:07
S1: it as discombobulating. It's downright humiliating. You've written that for many,
4:11
S1: it's actually a humiliation ritual. So tell us, I guess,
4:14
S1: how humiliating are we talking about?
4:16
S3: Well, I think I feel as if for older people
4:19
S3: it's particularly humiliating, particularly if you've got decades of experience.
4:24
S3: And I mean, it's it's natural to think, well, look
4:27
S3: what I've done. Look who I am. Don't you know
4:29
S3: who I am? Don't you know what my successes are?
4:32
S3: And all of a sudden I'm being, you know, greeted
4:34
S3: with this, this computer talking to me and eliminating me effectively. It's,
4:39
S3: I guess, another arm of ageism for a lot of
4:42
S3: people when they have had distinguished careers or even if
4:45
S3: they've had moderately successful careers, it's like, well, where is
4:50
S3: the respect for me as a person, and that process
4:53
S3: really seems to eliminate that. Whereas, I mean, in the past, yes,
4:57
S3: there could always be disrespectful recruiters and HR people or interviewers,
5:01
S3: but the fact that there was a human mediated process
5:05
S3: probably made it a little bit less humiliating.
5:09
S1: That's right. And you spoke to another woman, I believe
5:11
S1: she's Perth based in her early 50s. Phoebe, who said
5:14
S1: to you, I promised myself I wouldn't cry when I
5:17
S1: was talking to you, but it's really hard to keep going.
5:19
S1: I won't lie. You feel like such a loser, and
5:21
S1: I think anybody can sort of relate to that feeling
5:24
S1: in general, let alone in this sort of recruitment process.
5:27
S1: But is it just older people who are feeling humiliated
5:29
S1: by this and being weeded out by the AI platforms?
5:32
S1: Or is it other people?
5:33
S3: No. As it happened, I mostly spoke to people who
5:36
S3: were 40 plus, but in fact, I've had quite a
5:39
S3: few emails in response to the article since it was
5:41
S3: published on Saturday. And I heard from one woman who
5:44
S3: wrote to me about her two sons, both of whom
5:47
S3: were highly successful at school. One was the dux at
5:51
S3: his school and they did degrees in Stem and they've
5:56
S3: done extremely well in their studies. Neither of them can
5:59
S3: get a job. But you know, the mother is and
6:02
S3: her husband are just devastated, just feeling for their sons
6:06
S3: who've worked so hard all their lives. And my understanding,
6:08
S3: they're they're in their 20s. But, you know, these are
6:11
S3: the future of our country that we've, you know, we've
6:14
S3: invested in their education, their taxpayers invested in their education.
6:18
S3: The parents seem to have done everything right to sort
6:21
S3: of guide them on their way. And something within that
6:25
S3: AI process seems to be eliminating them from contention.
6:29
S1: And one category of people, I thought this was quite
6:31
S1: interesting because it's a broad category of people who seem
6:33
S1: to be, I don't know if discriminated against AI is
6:36
S1: the correct term, but basically being weeded out is people
6:39
S1: who have a varied career and life experiences and have
6:42
S1: built a wide array of skills, rather than a specific
6:44
S1: sort of highly honed few skills. Now, I would imagine
6:48
S1: a lot of those people who have, you know, lots
6:50
S1: of different transferable skills, they perhaps could do a very
6:52
S1: good job in whatever job they're applying for, but they
6:55
S1: are really finding it hard with this AI process. Is
6:58
S1: that right?
6:59
S3: Yeah. And I can count myself in this number as
7:02
S3: a middle aged journalist with 35 plus years of experience
7:06
S3: with no desire to retire, huge amounts still left to give.
7:09
S3: It's commonly talked about the decline in media and mainstream
7:14
S3: legacy media, the sort of jobs that I would have
7:17
S3: the ability to do, the capacity to do. I'm obviously
7:20
S3: not qualified because I've been knocked out of a few,
7:23
S3: but they call for extremely specific sets of skills. Stakeholder
7:27
S3: engagement and audience metrics, and social media demonstrated capacity to
7:34
S3: build audiences with social media. I can't put those exact
7:37
S3: keywords in any application, even though I have all of
7:41
S3: those skills. I haven't specifically worked in those areas, but
7:46
S3: it's second nature to me and what I do. So
7:49
S3: that's an example of, yes, I've got highly transferable skills,
7:52
S3: but not deeply specific ones that relate to the sort
7:56
S3: of jobs that are being advertised in media communications sort
8:00
S3: of sphere at the moment.
8:02
S1: And I know that in the article, this is Amanda
8:04
S1: Bollinger speaking about, I believe she's 48, and she said
8:07
S1: that while she looks for work, she's surviving on friends love,
8:10
S1: you know, loyal hairdressing clients and Airbnb cleaning work. So
8:14
S1: I guess how many Australians is this now happening to? Like,
8:18
S1: how common is it now for Australian companies to be
8:20
S1: using these AI platforms to weed through their job applicants?
8:24
S3: I think it's in excess of 80% of companies. I
8:27
S3: heard 90% of companies internationally as one figure. That is
8:31
S3: quite credible. Huge numbers. But what's impossible to know is
8:34
S3: how many people are actually affected and how many people
8:37
S3: are being demoralized and left on the shelf, if you like,
8:41
S3: when they're looking for work or retiring early, or going
8:46
S3: on to welfare payments or becoming homeless. You know, we do.
8:50
S3: Those sorts of stats are just impossible to find.
8:58
S1: After the break.
8:59
S3: And one person I interviewed said, it'll just be lots
9:01
S3: of little bots yelling at each other down the track.
9:05
S3: That's what the future of job hunting and recruitment looks like.
9:13
S1: And so the people or the companies who are actually
9:16
S1: using these AI hiring platforms in their recruitment process, what
9:19
S1: do they say is the upside to it?
9:22
S3: Well, there's a couple of tangential things here. As a result, obviously,
9:27
S3: normal job applicants are starting to use AI now for
9:30
S3: their applications. And anyone who's done a job application in
9:33
S3: the past decade, the old fashioned way, knows that it's
9:36
S3: intensely time consuming to do a good quality application because
9:40
S3: you need to retool your resume or CV to fit
9:43
S3: that particular job and emphasize and highlight and draw out
9:46
S3: the particular skills that are relevant to that job. Writing
9:49
S3: a cover letter. Oh my God, I've sweated for days
9:51
S3: over one cover letter. Just trying to get the nuance
9:54
S3: right that my personality right. It's so time consuming to
9:57
S3: do job applications. Whereas if you use ChatGPT or Gemini
10:01
S3: or Perplexity or any of these AI systems out there
10:04
S3: that so many of us are using these days, you
10:07
S3: can have your application done in, you know, 15 minutes
10:10
S3: or less. So recruiters I've interviewed say that there has
10:14
S3: been a deluge. It's increasing. So where once they might
10:18
S3: have got, you know, 100 applications for a position, they're
10:20
S3: getting thousands because it's so easy for people to send
10:24
S3: an application in. So you can see in that situation
10:27
S3: where an AI system that does an initial sort and
10:31
S3: filter is saving them an enormous amount of time, and
10:35
S3: therefore money. And they don't need to read through applications
10:39
S3: that are clearly dross.
10:40
S1: Yes. Interesting. And so you also wrote that recruiters widely
10:45
S1: acknowledged that these AI run systems can make poor decisions.
10:49
S1: So walk me through this. Like, how are the companies,
10:52
S1: I guess, being disadvantaged perhaps when the AI systems go wrong,
10:56
S1: I guess.
10:56
S3: Well, I mean, for a start, in terms of missing
10:59
S3: people who might be the most perfect person for the role,
11:03
S3: but whose skills don't immediately scream a match. For example, Amanda,
11:08
S3: the lovely, kind hairdresser with the psychology degree, she might
11:12
S3: have been the best candidate for some of the roles
11:15
S3: she applied for, but those AI systems are just completely
11:18
S3: ignoring her. Phoebe is another one. She is a major,
11:21
S3: major international player in her area, hugely successful woman in
11:27
S3: her career. And obviously she is for whatever reason. Maybe
11:32
S3: it's her age because she's now, I think, heading towards 50.
11:35
S3: She's been eliminated. Not even getting a look in. The
11:39
S3: CEO and founder of a really major executive search and
11:43
S3: advisory firm told me that just so often, the ideal
11:47
S3: candidates are just being who would be who are no
11:50
S3: brainer candidates are just being rejected out of the box
11:53
S3: because of these systems. But then there's the other things
11:55
S3: that academics are increasingly concerned about is the biases embedded
11:59
S3: in these systems. They can tell, you know, particularly say,
12:02
S3: if you've just done one of these video recorded interviews
12:05
S3: with your blank screen, it knows what ethnicity you might be,
12:09
S3: it might know if you might have a speech impediment
12:13
S3: immediately that's apparent. It can immediately tell you how old
12:16
S3: you are. And also the systems can be customized so
12:21
S3: that biases are built into it. I mean, there's actually
12:25
S3: been lawsuits in the states where companies have been fined
12:28
S3: for the biases that their platforms have embedded in them.
12:31
S3: But a platform could say, we don't want anyone who's
12:34
S3: black or we don't want anyone who's over 50 or
12:37
S3: we don't want women for this role. There was one
12:40
S3: example that I heard where a platform can embed exactly
12:44
S3: what companies it wants in someone's CV. So you, you
12:49
S3: might have worked for, say, Toyota and be applying for
12:51
S3: a different car company. And that car company might say,
12:54
S3: we don't want anyone that's worked for Toyota. We only
12:56
S3: want people that have worked for Hyundai or Bentley. No
12:59
S3: one will ever know what biases are embedded in these
13:02
S3: system in a really systematic way. Now, of course, human
13:05
S3: recruiters have always had biases. We know that, you know,
13:08
S3: that's that's not different. But this seems to be so
13:12
S3: much more malevolent and on such a mass scale as well.
13:16
S4: So tell me what's happening on those who are.
13:18
S1: Looking for jobs. Because you mentioned before about how people
13:20
S1: are now using ChatGPT and Gemini and so forth to
13:23
S1: sort of essentially speed up the process of applying for jobs.
13:26
S1: But you also wrote about people using those systems to
13:30
S1: build personal AI, job hunting agents to automate multiple elements
13:35
S1: of the dismal process. So tell us. I guess what these,
13:38
S1: you know, AI agents are doing for people and and
13:41
S1: whether it's working.
13:42
S3: I think the jury would be out still on whether
13:44
S3: it's working for people. But in the course of my research,
13:47
S3: I came upon a fellow in the US who works
13:50
S3: for Google actually, and has a side hustle writing, um,
13:54
S3: a Substack, um, about which he says he wants to
13:58
S3: democratize job hunting for everybody. And he's so techie, like,
14:04
S3: it's so much beyond my capacity to understand. And in
14:07
S3: his Substack, he takes people through the steps, all of
14:10
S3: the tech steps to do it. And you don't actually
14:12
S3: need a coding experience to, to do what he's proposing.
14:16
S3: But basically you can build with a number of steps,
14:19
S3: a bot, I guess, a little thing in the cloud
14:23
S3: that zips around and searches job boards every day for you. Well,
14:28
S3: you can sleep in or go off and have your
14:30
S3: coffee and the bot will do your. The searching of
14:32
S3: job boards for you. It will pull down, say, 3
14:35
S3: or 4 five jobs. I'm just pulling that number out
14:39
S3: of the air that you might be suitable for. It
14:41
S3: will then retool your CV to match that particular position
14:45
S3: and its position description, and then it will write your
14:47
S3: cover letter. And then when you've come strolled in from
14:50
S3: having your lovely coffee out in the sun, you go
14:52
S3: into your emails or into your digital files in one,
14:55
S3: one way or another, you check out what's found for you.
14:59
S3: You go, yep, I think that sounds a good job.
15:02
S3: Read the cover letter, check over the resume and off
15:05
S3: it goes. And you can send five applications a day
15:08
S3: for jobs or more or however or few or whatever
15:11
S3: you want. But effectively, it's taking out the grunt work
15:14
S3: of searching for for a job on, you know, there's
15:17
S3: multiple platforms, obviously, where there are positions advertised. And one
15:21
S3: person I interviewed said it'll just be lots of little
15:24
S3: bots yelling at each other down the track. That's what
15:27
S3: the future of job hunting and recruitment looks like.
15:31
S1: And so you have spoken to tons of people for
15:34
S1: this feature. Is there any way for a human person
15:37
S1: to game this system? You know, like, did you come
15:40
S1: out thinking, okay, you know, when I next want to
15:43
S1: get a job, I'm going to do X, Y, and Z.
15:46
S3: One person said the CV is dead. But I would
15:48
S3: say that that's not quite the case yet, given job
15:51
S3: ads are still calling for them. I think the only
15:54
S3: way to gain the system, and this is after doing
15:56
S3: two very large stories on this subject, is to take
15:59
S3: the advice of most recruiters and to stop applying for
16:02
S3: jobs you see advertised altogether, and just network, network, network, network, network,
16:08
S3: which is just the pits really, isn't it? But, you know,
16:11
S3: people talk about you've got to have one coffee a week,
16:14
S3: two coffees a week with people that could help you
16:17
S3: in your job search. And there are all sorts of
16:20
S3: strategies that recruiters and job coaches advise. You know, you
16:23
S3: don't go and say, have you got a job for me?
16:25
S3: You don't sit down at your coffee and sort of
16:28
S3: what can you give me? It's a more circuitous kind
16:31
S3: of conversation where you, you sort of slip in, I'd
16:34
S3: love your advice. I'm thinking of, you know, maybe moving
16:36
S3: into this sector or into this. I'd love to work
16:39
S3: in this company and I'd love your thoughts on that. And, um,
16:43
S3: and what that does, even that circuitous conversation, it puts
16:46
S3: you in someone's head. So if an opportunity opens up,
16:50
S3: if you're lucky, maybe there already is an opportunity at
16:53
S3: the company or that person knows an opportunity or is hiring.
16:56
S3: But even if there's not, that conversation puts you in
17:00
S3: their mind and they might go down the track if
17:02
S3: something comes up. Oh, I had a chat with Samantha.
17:05
S3: She would be perfect for that. I'm going to contact
17:07
S3: her and tell her to put her hat in the ring,
17:09
S3: and I'll make sure her application is stewarded through the
17:11
S3: process so it gets in front of the right people.
17:14
S3: That to me, is the only way I can see
17:16
S3: that you're going to game this system. And it's absolutely
17:19
S3: exhausting and I mean as soul destroying as well, because,
17:23
S3: you know, you could do an awful lot of coffees
17:25
S3: and not get not get anywhere.
17:27
S1: True. Although I guess at least you're doing it with
17:30
S1: people I don't know.
17:31
S3: Is that potentially. Well, I guess it depends on how
17:33
S3: extroverted you are and how much. Yes, I was. And
17:36
S3: if you're a misanthrope or you don't like people very much.
17:39
S1: Then it's true. I mean, how many diamonds in the
17:41
S1: rough or just absolutely excellent candidates out there are just
17:44
S1: not coffee date sort of people? Right. I guess that
17:46
S1: that's almost like the pointy end, isn't it, of of
17:49
S1: your future. Because that's what I got at the end
17:51
S1: of it was, you know, you said, this is your writing.
17:53
S1: You said if it's a small shaft of light, it's
17:55
S1: the experts I speak to, to tell me that the
17:57
S1: more AI systems are used, the more humans will be needed.
18:01
S1: And that, yes, you know, meeting face to face networking
18:04
S1: is still the best way to get a job.
18:06
S3: That's certainly what I have learned from from this research.
18:10
S1: Okay, so what I'm getting from this, Stephanie, is that
18:12
S1: you and I and everybody else perhaps listening should be
18:14
S1: very nice to all the introverts in our lives, you know,
18:18
S1: because this is what they're in for.
18:20
S3: Absolutely. And I mean, there are a lot of people,
18:22
S3: even the extroverts out there struggling as well. So it's
18:25
S3: a main street. It's yeah, I really think it's a
18:30
S3: tremendously sad place that we've sort of ended up at.
18:34
S1: Wow. Well, we're so lucky that you looked into it
18:36
S1: and you did such a deep dive into it. So
18:38
S1: thank you so much, Stephanie, for your time.
18:40
S3: You're welcome.
18:52
S1: In other news, today, after its own staff were unable
18:55
S1: to identify the bank as responsible for annual deductions when
18:59
S1: responding to queries from a customer who had been charged
19:02
S1: 13 years of fees for a card they never activated.
19:06
S1: Commbank is considering a call to more clearly name its
19:09
S1: fees on statements. A decision to green light, the Eastern
19:14
S1: Seaboard's largest wind farm in more than two years, has
19:18
S1: provided a much needed boost to Australia's renewable energy ambitions.
19:22
S1: Following a protracted investment drought that has stalled the rollout
19:26
S1: of new wind turbines and put the nation's 2030 climate
19:29
S1: targets under threat. And according to the Insurance Council of Australia,
19:35
S1: solving the housing shortage faces an overlooked competitor. The $11.3
19:40
S1: billion spent rebuilding and repairing homes damaged by extreme weather
19:46
S1: over the past four and a half years. You can
19:48
S1: read more at the age.com.au or smh.com.au. Today's episode was
19:55
S1: produced by Josh towers. Our executive producer is Tammy Mills,
20:00
S1: and our podcasts are overseen by Lisa Muxworthy and Tom McKendrick.
20:04
S1: If you like our show, follow the Morning Edition and
20:07
S1: leave a review for us on Apple or Spotify. Thanks
20:11
S1: for listening.