by Peggy Fiebig
The use of algorithms, including artificial intelligence, is playing an increasingly important role in searching for and recruiting new staff. Companies hope algorithms will help them make more accurate decisions in the job application process. But how does AI-based staff recruiting actually work? What options are available now?
For job applications, these programs perform language analyses to identify personality traits.
Other programs, such as profiling and matching software, use algorithms to search career websites for suitable candidates.
Job requirements must be defined in detail to ensure accurate results.
The results of AI-based staff selection should serve only as a decision-making aid because software cannot substitute for empathy and an understanding of human nature.
It sounds like a scene from a science fiction film: Instead of meeting with a pleasant member of the HR staff, the job candidate’s interview is conducted by a computer, which poses the questions and evaluates the answers on the spot. The computer then takes the decision: pass or fail. It was a quick and – above all – objective process with no human prejudice or gut feelings involved. This may sound like something in our distant future, but the fact is that first steps in this direction are already being taken.
Relationship between speech patterns and personality structure
First example: Aachen-based Precire Technologies has developed a program that uses language analysis to identify specific personality traits of job applicants. All it takes is a fifteen-minute interview to draw conclusions on personal characteristics, sources of motivation and attitudes based on speech patterns. Co-founder and CIO of Precire Technologies, Christian Greb, is a trained psychologist and previously worked at a management consultancy firm that addressed the topics of motivational psychology and personality psychology. Six years ago, he says, efforts began to develop a tool.
“You are HOW you speak”
It all started when they realised that communication and language convey much more than just the factual content of what is being said. It is not only: “You are WHAT you say”, but also: “You are HOW you say it”. With the support of researchers at the RWTH Aachen University, Greb explains, it was then possible to use algorithms to correlate speech patterns and personality structures. The research pool included roughly 5,500 people and data was collected during personal interviews and taken from psychological personality analyses. After a year and a half of research, they found that the algorithms developed had an accuracy rate of no less than 85 to 90 per cent. In 2013, Precire was launched on the market and has since been used not only in recruiting processes, but also as the basis for internal company staff development measures and other speech and text optimisation initiatives.
A detailed job requirements profile is a must
How exactly does this work? The first step for any company involves creating a detailed target profile: What skills should the future employee bring to the job and what should they be like? Does the position require a leader or a team player? More of an enthusiastic optimist or perhaps a cautious sceptic? The more precisely the requirements are defined, the better the results that can be achieved, explains Christian Greb, CIO of Precire Technologies. Job applicants are then invited to participate in a short interview conducted on an Internet platform. This brief discussion can but does not have to focus on professional content because the fact is: It is not about what the candidates say in response to questions, but how they say it. Which words do they use particularly frequently and how long are the sentences they build? How many and how long are breaks in speech? From the several thousand parameters stored in the program, the software aims to deduce, for example, whether someone is more authoritative or adaptable, or how willing they are to take risks. Overall, the Precire algorithm can measure 42 dimensions of personality. The company then receives an evaluation based on their job requirements to help them during the decision-making process. However, Christian Greb is quick to stress that under no circumstances should anyone be hired based solely on Precire’s findings. Precire cannot and should not make autonomous decisions. After all, only a human being capable of empathy and insight into human nature can really determine if a potential candidate is a good match for the company.
Application filter software like Precire is only one form of automated recruiting. Its main goal is to serve as a preliminary applicant selection tool. There are also tools that can actively search for candidates.
Automated search of career platforms
Second example: LogOn is a profiling and matching software. Using algorithms, LogOn searches career websites such as Xing and LinkedIn for suitable candidates. Again, the first step is to create an accurate job description, which is then used to search through thousands of potential candidate profiles for the best matches. The process focuses on the context as a whole and does not depend on individual – and possibly coincidental – terms used. In this way, explains founder and managing director Peter Kolb of LogOn Consulting, gaps or discrepancies in profiles are assessed very differently than if a person conducted the analysis. “By focusing on the big picture, taking into account the overall skill-set and experience, we come across candidates who would slip through the fingers of a recruiter only looking for a particular job title.” Another advantage: Companies no longer have to place expensive job ads on a wide variety of platforms to cover all bases – and then invest the time and effort in rummaging through hundreds of applications. The tool also provides information on how best to recruit candidates it has recommended as suitable.
Photos: © shutterstock.com/Zenzen