top of page
Search
  • th9934

Generative AI Implementation


From Shortlist, Thursday, 02 November 2023 2:23pm

One year on since the launch of ChatGPT, many recruiters

remain 'stuck' at the trialling phase, with upskilling time and

privacy concerns the main barriers to fully embracing

generative AI.

At a recent Captain's Table briefing hosted by Navigator

Consulting MD Tony Hall, the majority of delegates said

they have experimented with generative AI; a few are

comfortable using it, but none deemed themselves 'expert'.

This reflects the broader Australian market, where 'early

adopters', or those who are now comfortable using

generative AI, make up only a quarter of employees,

JobAdder CEO Martin Herbst told the briefing.

But as "one of the most human-based roles out there",

recruiters are missing a big opportunity if they are not

maximising their use of AI and benefiting from "AI companionship", Herbst says.

And enlisting the help of these early adopters might be a good way to socialise AI in the company, he

adds.

"You identify them as super users. These are the people who can actually be champions. You give them

that sponsorship, that support... They can be the trailblazers that try some things out and then share back

with the rest of your team to bring them on that journey."

Current usage

Among recruiters in attendance who have trialled or are currently embracing generative AI, the most

popular use is for writing candidate summaries. Other top mentions include creating business

development pitches, job descriptions and job ads.

One recruiter who is "probably using it on a daily basis" says it's particularly valuable for writing candidate

summaries.

"Some of the roles I'm working on [are] like your $150k-plus candidates. So they're very, very chunky

backgrounds, which you do need to formulate into something very professional but a small snapshot. I find

it very, very helpful with that," she says.

Another recruiter adds: "I like it because you can pull the main details, that profile page of a candidate,

and tell them to break it down into three top paragraphs.

"It can be quite clunky though. That's what I don't like, they make it quite complex in how they're

describing a candidate. They'll use very complex words that I then need to prompt again [to be] 'simpler' or

'more succinct'," she says.

"They tend to overcommunicate what you're trying to say, but now I use it quite religiously with summaries,

to be honest."

In using ChatGPT for very technical job descriptions, one recruiter describes the process as "perfect",

noting that what would normally take two weeks to put together can be done in two minutes.


Tony Hall, Navigator Consulting

"I'm at the point where I could say 'summarise this and in the third-person [voice]'. I can give it the right

prompts... It's just taken time to be able to figure out what the right way is to prompt it to get what you want

out of it, but I'm still experimenting."

Hall mentions another recruiter who runs all his emails

through ChatGPT to check for grammatical errors and typos.

"He said he had to write a very difficult redundancy email

last week and ChatGPT phrased it in such a better manner

and [with] a more human element.

"Ironically, it's a machine doing it, but the human factor he

couldn't get across in a redundancy email that went with the

meeting he had with the person, he said was just done

brilliantly."

Why recruiters have stopped using it

Learning how to prompt effectively has been a common

challenge among recruiters trying out generative AI this year,

delegates agree.

"I stopped using it... probably because I didn't work hard enough at the prompts to be fair," says one

recruiter.

"And then the other side of it was, a lot of the times when I'm speaking to a client, I have the interview

locked in before I send over the CV... So I find there wasn't necessarily that need for me to write those

extensive candidate summaries because of the relationships we've actually had to work with different

businesses."

Herbst also mentions the significant time it takes training ChatGPT to speak in recruiters' own unique

style. If it's not producing content or output that recruiters feel is "in their own voice", there might not be

enough value to keep using it, he says.

No way to retract personal information

Meanwhile, the biggest sticking point with recruiters, still, is regarding data and privacy.

"I've tried [ChatGPT]. The more you prompt, the better it gets," one delegate says. "[But] we won't do

candidate summaries in ChatGPT at this point because we don't feel like we've resolved the Australian

privacy legislation aspect of that."

Further, "there's no sort of way that our candidates can retract their personal information from ChatGPT

[and] we're not explicitly asking them 'can we put their data into that platform?'", she says.

Another says: "If I was to do a candidate summary, it would be more general bullet points about them and

then I get it to formulate and make it sound better for the client application. But I would never put personal

details in there, or company names.

"You do a little bit of work behind the scenes to make sure nothing does go in there, but we've definitely

got pretty strict policies in place that got rolled out across the company."

bottom of page