Artificial-intelligence-based hiring tools are already transforming the recruitment process by allowing businesses to vastly speed up the time it takes to identify top talent. With algorithms able to scour applications databases in the fraction of a time it would take a human hiring manager, AI-assisted hiring has the potential not only to give precious time back to businesses, but also draw in candidates from wider and more diverse talent pools.
AI-assisted hiring is also posited as a potential solution for – reducing human bias whether subconscious or otherwise – in the hiring process.
SEE: Robotic process automation: A cheat sheet (free PDF) (TechRepublic)
US company Harqen has been offering hiring technologies to some of the world’s biggest companies for years, partnering with the likes of Walmart, FedEx and American Airlines to streamline and improve their hiring processes. Originating as an on-demand interviewing provider, the company has now expanded into AI with a new platform that it says offers a more dependable and bias-free means of matching employers with employees.
The solution, simply called the Harqen Machine Learning Platform, analyses candidate’s answers to interview questions and assesses the type of words and language used in their responses. According to Harqen, this allows it to put together a profile of psychological traits that can be used to help determine a candidate’s suitability for a role.
Combined with a resume analysis, which provides a more straightforward determiner of whether a candidate’s professional and educational background fits with the requirements of the job, Harqen says its machine-learning platform is capable of making the same hiring decision as human recruiters 95% of the time. In one campaign that assessed approximately 3,500 job applications with “a very large US diagnostic firm” in early 2020, Harqen’s machine-learning platform successfully predicted 2,193 of the candidate applications that were accepted, and 1,292 that were declined.
Key to Harqen’s offering is what the company’s chief technology officer Mark Unak describes as the platform’s linguistic analysis, which can identify word clusters that are specific to certain job types but also offers a personality analysis based on the so-called “big five” traits, also known as the OCEAN model (openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism), which can help hiring managers determine a candidate’s enthusiasm for the position.
“We have a dictionary of terms where most positive words are ranked as a +5 and most negative words are ranked as a -5, so we can determine how enthusiastic you are in the answers that you’re giving,” Unak tells TechRepublic.
“We can also use a linguistic analysis to analyse the grammar,” he adds, noting that about 60% of our vocabulary consists of just 80 words. Those are the pronouns, the propositions, the articles and the intransigent verbs. “The remaining 10,000 words in the English language fill in that 40%. By the analysis of how you use that, we can get a psychological trait analysis.”
According to Unak, using a machine-learning system that determines a candidate’s suitability based on linguistic analysis is a more accurate and impartial method than those that rely on facial-scanning or vocal-inflection algorithms. Such machine-learning techniques within hiring are on the rise and are increasingly being adopted by major companies around the world.
“That’s kind of problematic,” says Unak. “Not everybody expresses emotions in the same way, with the same facial expressions, and not everybody expresses the same emotion that’s expected. Different cultures and different races might have different problems in expressing those facial expressions and having the computer recognise them.”
SEE: Diversity and Inclusion policy (TechRepublic Premium)
By only analysing the linguistic content that has been transcribed from recorded interviews, Harqen’s algorithm never factors in appearance, facial expressions, or other self-reported personality traits that could be unreliable. Unak says the company will also retrain its models on a regular basis as new data comes in, which will help ensure that algorithms don’t get stuck in their old ways if candidates begin giving new answers to questions that are equally relevant.
“If our customer evolves and they start to hire people who are either more diverse, or come up with different answers to the questions that are just as relevant, our models will pick that up,” Unak adds.
Diversity – whether based on gender, race, age or otherwise –has been show to play a significant role in the success or failure of workplace productivity and collaboration. Whether AI-based hiring tools can help here remains to be seen, and ultimately depends on whether they can be implemented in a fair and impartial way.
Beyond diversity, Harqen is exploring how its machine-learning tool could help businesses get the best return on investment form their hiring choices. The magic word here is delayed gratification: the ability to accurately identify employees who can resist the temptation for immediate rewards and instead persevere for an even greater payoff in the future.
“It’s grit, it’s persistence, it’s the ability to imagine a future and it’s the ability to develop and execute a plan to get there,” says Unak. “Isn’t that what hope and delayed gratification mean? I hope for a better future, I can imagine it, my hope is realistic and that there’s a plan or a way to get there, and I’m going to work towards it.”