The job interview. Proving your usefulness and value in front of someone is a daunting prospect, yet every year, thousands and tens of thousands of intense, life-changing applications are submitted to a single company for the chance to get an interview, and the majority of them end up with the bitter pill of rejection.
Even if you do manage to get an interview, the struggle continues. You have to pray that you don't get caught in the crossfire of an overly demanding interviewer who puts you on a pedestal, puts you on the spot with pressure questions, or insults your personality.
How can we give more people the opportunity to interview and get a closer look at each person's potential, which is hard to capture from a few lines of specs on paper? How can we evaluate them fairly so that they don't fall victim to interviewer bias? This is where AI interviewing steps in, offering a potential solution to these concerns. How can we avoid wasting time for both interviewers and candidates?
This is where AI interviewing steps in. It's no secret that many companies have embraced AI interviews for recruitment in recent years. However, there are valid concerns from the applicant's point of view about the level of objective evaluation criteria.
My name is Geunsik Yuk, and I am in charge of the HR business at Genesis Labs. Genesis Labs develops and services HR (Human Resources), mental health, and content platform solutions based on AI technology. Our representative product in the HR field is ViewerHR, an AI video interview solution that I will discuss today.
Before joining Genesis Labs in early 2022, I had about 18 years of HR experience at LG Electronics and LG U+. I was also the head of the recruitment team at the first customer (LG U+) that utilized ViewinterHR for recruitment.
ViewinterHR is powered by deep learning technology, which is how AI learns to watch and label (evaluate) hundreds of thousands of interview videos and answers from recruiting and interviewing experts. It evaluates the content of answers and non-verbal elements (tone of voice, gestures, etc.) just like a human interviewer. We are constantly working to minimize bias in the evaluation and improve the performance of the AI.
We use speech-to-text (STT) technology to analyze and transcribe interview responses and text-to-speech (TTS) technology to convert text-to-speech by synthesizing human speech. Genesis Lab has registered and applied for 67 domestic and international technology patents to provide better AI interview solutions.
Deep learning methods need to learn insights from humans. ViewinterHR has secured over 100 of the best labellers (evaluators) in Korea, including former HR executives of large companies, renowned HR and organizational management experts, professors of industrial psychology, and professional interviewers from public institutions. The AI learns from the data they evaluate after watching interview videos and produces interview evaluation results.
However, labellers are human, so not all ratings are perfectly consistent, and interviewer fatigue can even slightly affect the content of the ratings. Technology compensates for human resource limitations. By analysing and monitoring the data, we can thoroughly investigate and remove bias.
We are the only HR company in Korea to be certified by the Korea Telecommunications Technology Association (TTA) for AI reliability assessment, and we have won awards from the Ministry of Science and ICT, including the 'AI Reliability Quality Award' and the 'Korea Artificial Intelligence Award'.
We evaluate both the hard and soft skills of candidates. First, hard skills are also known as "job skills." Based on the content of the answers, we evaluate the candidate's significant knowledge and suitability for the job. You might ask questions about past job-related experience, for example.
Soft skills (common competencies) are based on the video, audio, and content of the answers to see if the candidate fits the organization's requirements and has the creativity, enthusiasm, and responsibility to match the talent profile.
I know that other AI interviews have the same questions for entry-level and experienced candidates, but we have different questions. The skills and knowledge required by various industries and job roles differ, so we optimized questions to suit each characteristic. For example, if you're interviewing for a sales position, you might ask questions about customer negotiation skills.
Yes, some of our customers create the interview questions themselves, and we collect and evaluate candidate data with them when designing and training deep-learning AI. We also customize behavioural indicators of talents or core competencies based on their input.
ViewerHR, an interactive AI video interview solution
ViewerHR, an interactive AI video interview solution developed by Genesis Labs (Photo courtesy of Genesis Labs)
Based on the AI's learned data, it evaluates the applicant's video, voice, and answers. The absolute scores are then compared to the relative scores to create comparative scores, ratings, and grades. It's the same way as most HR teams that conduct in-person interviews. The AI interview report, candidates' interview videos, and transcripts are all delivered to the company. How they use these materials and the logic behind the pass/fail decisions varies from customer to customer.
Suppose you need an AI interview process in the middle and go straight from the paperwork to the in-person interview. In that case, companies that get thousands and tens of thousands of applications will only accept a few people because they can't get everyone in for an interview. If you have 10,000 people applying, 9,000 won't get an interview.
But there are many people who have skills that don't show up on paper, who are much more compelling when they sell themselves orally rather than on paper, and it's a shame for companies that they can't give them a chance due to resource limitations. AI interviewing allows you to dramatically increase the number of candidates you can interview because you don't need human interviewers to interview them.
The second advantage is that the selection process is fairer. Depending on the interviewer's mood and personality, in-person interviews can be a hit or miss. AI interviews minimise variables and ensure every candidate receives a fair assessment based on objective metrics.
It's also a massive benefit for candidates to be able to prepare for interviews on their own time and space, as they often have to travel long distances and juggle schedules to attend in-person interviews. AI interviews can be conducted at any time of the day, night, weekday, or weekend, as long as they are within the candidate's timeframe and can be accessed on mobile.
ViewinterHR's AI interview is built with evaluation logic by deeply learning insights from HR and recruitment experts (humans), so it's best to be as natural as you would be in a face-to-face interview. It's hard to make a good impression face-to-face with an interviewer if you're trying to make a conversation. Think of it as a person, not an AI, on the other side of the screen, and be calm but sincere in your answers, and you'll get good results.
At that time, we first used AI interview solutions for new recruitment, which involved conducting personality tests between the screening process and the first interview. However, I thought it would be much more efficient if there was a tool that could screen candidates in the middle of the process. The timing was right because digital transformation (DX) was a hot topic in all areas of business, and recruitment is the most suitable area for DX innovation in HR.
However, it's unreasonable for even the most prominent companies to develop their AI interview solution, so I met with several companies that use the technology and decided that Genesis Labs was the closest to what I wanted. Genesis Labs' ViewerHR is the first interactive AI interview solution in Korea, and I found it attractive because it can be evaluated from the perspective of 'others'.
The area that needs to be further developed is interaction. In terms of the Q&A interaction, it's already very similar to a face-to-face interview, but we're not yet at the stage where we can give appropriate feedback to the candidate's answers and ask personalized follow-up questions on the fly, which is essential for deeper candidate qualification and a better hiring experience for the candidate. We're actively working on R&D and expect to be able to introduce this feature shortly.
We currently have over 150 customers using ViewinterHR for hiring. About 60% are private companies and 40% are public. Private companies include LG, Hyundai Motor Company, Hyundai Department Store Group, Hanwha Group, and Kookmin Bank. Public companies include Korea Water Resources Corporation, Gyeonggi Metropolitan Assembly, Korea Railroad Corporation, and the Army, Navy, Marine, and Air Force.
AI human technology has made it possible to have interactive interviews in a more relaxed and natural atmosphere. In the future, we can provide AI interview solutions that are as interactive as face-to-face interviews with human interviewers by advancing generative AI technology. It is expected that the business's future success will be determined by how all IT companies utilize and apply generative AI, not just in the AI interview field. Genesis Lab aims to provide companies and candidates with a more satisfying recruitment experience through generative AI technology.