The Power of AI for Recruitment

The rise of artificial intelligence is changing the game when it comes to finding the perfect job candidate for a role.

AI tools are being used by recruiters to automate tedious tasks such as screening resumes and candidate profiles, evaluating skills and qualifications, analyzing job postings, and even conducting initial interviews.

Not only can AI make the recruitment process faster, easier, and more efficient than ever, but some claim it can even reduce human bias by helping recruiters quickly identify qualified candidates from a larger pool of applicants

Let's take a look at some ways recruiters are making the most of artificial intelligence to streamline the hiring process.

Boolean searches use the terms “AND”, “OR”, and “NOT” to combine specific keywords or phrases to create a more precise search

Recruiters use them to narrow down online job postings and search for candidate profiles who have the relevant skills and experience for a position - even if they may not be looking for a job.

Sourcing and Boolean search expert Glen Cathey used ChatGPT's expertise to write Boolean search strings. He asked;

"Write a Boolean keyword search to find a resume of a software engineer with JavaScript, node, SQL, and financial services industry experience."

Output: JavaScript AND node AND SQL AND "financial services" AND (software OR engineer OR developer)

He noted that the output he got was basic and he was not too impressed. Although it may have been technically accurate, it may not have captured all the potential talent. Maybe a candidate represents "financial services" experience with a different word. He was impressed, however, that ChatGPT included the term "developer" as a synonym for "software engineer," even though he didn't specifically ask for it. So a more successful Boolean search would include many more keywords - perhaps a good task for ChatGPT.

Companies like Occupop, a recruitment software company, also found value in using ChatGPT to generate Boolean search strings. Their initial prompt was more descriptive;

"Create a Boolean search string for LinkedIn to identify a financial analyst with experience in large multi national financial institutions. The financial analyst should have experience in Fintech and be based in London. Also include alternative ways in which people may enter their experience as a financial analyst in their LinkedIn profiles."

But they found the output was particularly successful when adding an element of research into the input such as;

"Come up with the names of the top 50 nursing homes in the UK and put these into a string."

They got a long list of specific nursing homes in the desired area formatted in a Boolean search string (essentially the names of the nursing homes separated by "or") which could be helpful for recruiters who are looking for similar candidates within a particular industry.

Miles Jennings from recruiter.com also feels that ChatGPT is also a great help for creating job postings. He prompted;

"Write a compelling job description for a Java developer."

Did it work? Yes. It admittingly wasn't the most convincing job description but it listed enough qualifications to make it a good base to start with. He suggested improving it with;

“Now make that same job description more interesting” or “Now emphasize the need for J2EE design patterns in the job description.”

Then he wanted to adapt the formal job description that he just generated to make it easier to understand and possibly more attractive to potential candidates on different platforms.

“Now write a summary of that job description for social media, and make it really interesting and intelligent, meant for Java Developers.”

The output started with "Calling all Java Ninjas," which Jennings said nailed the way recruiters usually write on social media.

ChatGPT is also great at drafting cold outreach emails for engaging passive candidates. Occupop shared their prompt which, again, is quite detailed;

"I will give you a job description and you will write an email trying to engage the candidate. Begin with their potential interests in career growth, then present the job opportunity. Outline the benefits and use a convincing tone. Use industry terminology. Keep it short, do not use unnecessary words. Be friendly and human in tone. Use call-to-action, add a sense of urgency.”

They considered the output decent but noted it would need some minor tweaking before it was ready to post. Then, to expand on that, they also asked ChatGPT to generate multiple email subject lines to maximize the click rate.

"Create a list of convincing email subject lines to engage a candidate."

Again, the results were not award-worthy, but acceptable with a few adjustments.

Interviewing is one of the most important parts of the recruiting process and one that includes a bulk of the tedious tasks that AI tools can help with like preparing for the interview and summarizing it for the hiring managers.

So many recruiters across the internet seem to be excited about using ChatGPT to create interview questions. It seems to be especially useful when they are hired to search for a candidate in a field that requires specialized knowledge or skills in which they have no expertise.

For example, when recruiter Erondu Caleb was hired to recruit for a UX designer position, he turned to ChatGPT;

"Act as a recruiter interviewing a UX designer for a role. What questions would you ask in initial interview?"

And he was quickly given a list of 10 questions he could ask in an initial interview, saving him hours of research time.

Honeit co-founder Nick Livingston took this idea and ran with it. He created a hiring software that uses GPT-3 technology to enhance the interview process, from start to finish.

Honeit's idea is that interviews are full of valuable data. And recruiters spend a lot of time after an interview typing up a summary of that data that they can then share with the hiring manager, which is where the Honeit platform comes in.

To demonstrate how it works, they shared a use case from one of their users in a recent podcast. The recruiter, who specialized in the marketing field, was tasked with recruiting a Senior Software Engineer at LogRythm.

So they copied the job description and pasted it into the "AI Guide" in Honeit, which, like Caleb's example above, gave 10 targeted interview questions along with the best development skills from the posting listed as keywords.

The recruiter then goes on to conduct the interview with these questions through the Honeit platform. The Honeit technology captures recordings of the interview, creates a written transcript of the interview recordings, and then prompts GPT to extract data.

What the recruiter receives is a highlights page that includes the recordings of the interview questions that they

The data is parsed together in summaries on a highlights page for the recruiter. They can then share the recordings and summaries, via audio or text with people who were not on the call saving the recruiter from typing up an extensive summary and allowing hiring managers to make informed decisions based on summarized data rather than relying solely on recruiters' opinions which, as Honeit claims, reduces basis.

So in this example, the prompts are done within the Honeit technology when they ask GPT to summarize the extracted transcripts. The two Honeit podcast hosts called themselves "prompt engineers" and are understandably not releasing their prompts since they are considered trade secrets.

We've seen one argument for how AI in recruiting reduces human bias, but others argue that AI-powered recruitment processes can also lead to a replication of existing biases because AI is heavily trained on past data and there may not be enough diversity in that data.

And while laws around AI are still quite vague because it is a relatively new technology but labor laws are pretty concrete so companies need to be aware of the legal implications as well.

The US Equal Opportunity Employment Commission EEOC has previously investigated discrimination in job decision algorithms on several accounts and has since launched an initiative on artificial intelligence and algorithmic fairness.

Since the ethics behind using AI for the recruiting process is still highly debated, businesses should take steps to ensure their data sets are diverse and representative of the population they are trying to recruit from.

Ultimately, while AI-powered technologies have the potential to streamline the hiring process, it's not likely to replace the human touch that is needed to effectively assess a candidate's qualifications.

They should be used in conjunction with traditional hiring methods like two-way interviews, and alongside experienced human recruiters who understand how to get the most out of these tools without sacrificing the quality of hire or the candidate experience.