It has appeared in the fields of banks, foreign companies, fast-moving consumer goods, etc.AI Interviewer, is penetrating into more industries.
A fresh graduate applying for a hotel front desk customer service position told Tech Planet that although she had received a notice to go for an offline interview, she still needed to complete theAI InterviewAt the offline interview site, the interviewer would specifically ask her about her experience using AI interviews, "This hotel has just launched AI interviews, mainly for testing, and would like to know our feedback."
Source Note: The image is generated by AI, and the image is authorized by Midjourney
At present, the "golden March and silver April" recruitment season is coming to an end. According to data, the number of college graduates in 2024 is expected to be 11.79 million, an increase of 210,000 year-on-year. The competition in the job market is becoming more and more fierce, especially in industries with low acceptance ratios like Procter & Gamble's 5000:1. AI interviews have become an important method for companies to initially screen candidates.
A recent graduate who had just participated in five consecutive AI interviews in one day complained to Tech Planet that it was just like when they wanted to consult human customer service in their daily lives, they would always encounter AI customer service. Now if they want to get a spot for an offline interview, they need to pass the "invisible threshold" of AI first.
According to a report released by market research firm Resume Builde, 43% of companies are expected to introduce AI applications in recruitment by 2024. In this spring recruitment, AI interviews have become an indispensable presence, and many job seekers describe it as a "one-man show". They keep practicing in front of the screen to make their facial expressions not too stiff and to organize their language more logically, so as to accept or even please the "digital human" interviewer opposite.
Participating in dozens of AI interviews increases job search costs
At 10 a.m., sitting in front of the computer, Zhang Lu began her seventh AI interview. This time, she applied for a job at a large Internet company. Although the test page at the beginning showed that the network was good, when she actually started recording, the "network lag" prompt popped up more than ten times, constantly forcing her to return to the initial test page.
The questions asked by the AI interviewer rarely involved the professional knowledge of the position she applied for, but were relatively broad and applicable to all positions or industries in the entire company. The AI interviewer also asked intelligent follow-up questions, but this time the AI interviewer asked questions that she had already answered in her previous answers, which resulted in her repeating the questions three times throughout the interview.
Sometimes, she felt that the digital person on the other side was "arguing". "For example, I wrote on my resume that the customer satisfaction of a certain project reached 80%, and the AI interviewer would ask me why it was 80%?"
After answering the last question, Zhang Lu let out a long sigh, "There were so many unexpected situations. I felt like I was going crazy from this AI interview."
Like Zhang Lu, there are mixed reviews on social platforms about AI interviews, with many job seekers even calling them “torture” and “anti-human design.”
Some job seekers who participated in AI interviews for the first time encountered many problems. For example, when facing the camera, they suddenly became incoherent; they answered the questions in 40 seconds instead of five minutes, and started to make up stories in a hurry because they were worried that the time was too short. As a result, AI started to ask about these fabricated experiences. "I was stunned for a moment, and I didn't know whether to laugh or cry in front of the screen," a job seeker told Tech Planet.
AI interviewers are not a new concept. They have been used in companies such as Unilever, L'Oreal, and Coca-Cola since 2019. Job seekers usually receive AI interview information immediately after submitting their resumes. AI interviews usually have about 7 questions. After the interviewer asks a question, the interviewee will be given about 5 minutes to answer. The entire interview usually lasts 20 to 30 minutes.
What is now widely discussed again is that AI interviews are becoming more popular in more companies. When Tech Planet talked with many college students, they all said that they had had many AI interview experiences.
Caption: AI interview interface; Source: Xiaohongshu.
During her job search, Yang Yang went through more than a dozen AI interviews. According to her, AI interview questions from banks, state-owned enterprises and other units tend to be more traditional, while those from fast-moving consumer goods companies are more flexible, but they are basically similar to the "P&G Eight Questions", such as: In your past experiences, what aspects of your leadership skills can best be demonstrated?
During the AI interview, screenshots of questions are not allowed, as they will be displayed as violations. If too many screenshots are taken, the interview will be deemed invalid, Yang Yang said. Therefore, before the AI interview, everyone basically goes to major social platforms to search for various AI interview strategies and question bank summaries.
Every year during the spring recruitment, college graduates will submit a large number of resumes and participate in dozens or even hundreds of recruitments. In Yang Yang's opinion, if every company adopts AI interviews, it will increase the invisible job search costs. After all, preparing for an AI interview usually includes multiple processes such as searching for interview experience in advance, adjusting answers, putting on makeup, and meeting deadlines, and the result is often "no response" and no news. She would rather companies screen out a wave of candidates at the beginning rather than let every job seeker participate in AI interviews.
Yang Yang still remembers,FirstShe was nervous when she first participated in an AI interview, so she would focus on practicing her eyes, hoping to be more natural in front of the camera. But after participating in too many AI interviews, she gradually became numb and was too lazy to put on makeup. She also tried to let ChatGPT answer questions raised by AI interviews, focusing on "using magic to defeat magic."
Who are the AI interviewers?
The AI interviewers used by companies are usually provided by third-party agencies, and are mainly used in the initial screening stage between resume submission and offline interviews.
At present, the functions of AI interviewers on the market are relatively similar. They use different algorithms to analyze the interviewee's answers, facial expressions, body movements, and voice, and finally generate an evaluation report and display the ranking of all job seekers for the position. This has become an important factor in whether companies approve candidates for offline interviews.
The company can set whether to display the evaluation report to job seekers in real time in the background. However, according to the experience of many job seekers, most companies currently choose not to display it.
The person in charge of UFIDA Dayi's AI interview product told Tech Planet that the core of AI interview products is the construction of job competency models. The AI interview launched by UFIDA Dayi was jointly developed by HR experts, psychology experts, AI experts, etc. It has a set of question banks that can comprehensively examine the job competencies of candidates, and companies can also modify and adjust according to their own needs.
The evaluation report received by the company is based on the assessment dimensions of the "AI interviewer", which includes both external and explicit traits such as professional knowledge, expression ability, image and temperament, as well as more implicit information such as competence, mental health, personality, and values.
Companies can select several dimensions for key assessment based on job requirements. For example, for sales positions, candidates’ communication skills will be assessed, while for blue-collar positions, mental health, basic Mandarin expression, and physical flexibility will be assessed more.
More and more companies are choosing AI interviews, which is inseparable from the four words "cost reduction and efficiency improvement". Fast-moving consumer goods giant Unilever once released a set of data: through the AI system, 100,000 hours of interview time were saved, which can save 1 million recruitment costs each year.
The relevant person in charge of "Doris", the AI interview product of Liepin, told Tech Planet that AI interviews are currently mostly used in state-owned enterprises, banks, fast-moving consumer goods, catering and other industries, and are mostly used in scenarios such as campus recruitment, which can effectively improve the recruitment efficiency of enterprises.
For example, in a bank's campus recruitment, if only the traditional offline 5-minute quick interview is adopted, it will take 20 HRs 3 days to complete 5,000 interviews. If the multi-faceted AI digital officer Doris is used, 5,000 interviews can be completed in 48 hours, which is equivalent to one digital interviewer completing the work of 20 HRs in 3 days in 2 days.
Using AI interviews requires a certain cost, so it is more widely used in medium and large enterprises. An AI interview service provider consulted by Tech Planet said that one way is to charge according to the number of people, adopting a pre-charge model, with a single-use fee of 15 to 30 yuan. Enterprises can enjoy AI digital interviewers, evaluation reports, question banks, SMS and email batch notifications and other benefits.
Another charging model is to cooperate with enterprises on customized projects. Some enterprises will involve some value-added services, such as customized interviewer images, etc. For example, if you cooperate with a bank institution for spring campus recruitment, the fees usually range from tens of thousands to hundreds of thousands of yuan.
Is AI interview reliable?
"Will HR review the AI interview video?" and "Will I be eliminated by the algorithm?" are questions that many job seekers are concerned about.
At present, there will be special reminders in the AI interview notifications of many fast-moving consumer goods companies. Real interviewers will watch the AI interview videos and use them as one of the bases for offline interview invitations.
An HR who has been in the industry for many years told Tech Planet that companies use AI interviews for two purposes. One is to use it as an efficiency tool in large-scale recruitment, using AI to quickly locate candidates' rankings in the talent pool. The other is to serve as an auxiliary evaluation reference. If the recruitment volume is large, HR will not check each one one by one, but may conduct random checks and verify the accuracy of the AI evaluation.
As a graduate of the human resources major in 2024, Li Mang experienced a dramatic scene. During her internship this year, she was responsible for reviewing the evaluation reports and interview videos of AI interviewers for screening.
Li Mang works for a medium-sized company. She told Tech Planet that after the company sends out an AI interview notice, it will follow up on the results in a timely manner. After all, the cost of each use is at least a dozen yuan. Companies consider different indicators for different positions. For example, sales positions focus on goal-orientedness and execution, HR positions focus on affinity when expressing, and finance-related positions focus on logical ability.
In the background, Li MangFirstThe first time I saw the results of AI interviews from the perspective of HR, the interview performance of job seekers was quantified into specific indicators one by one, and even the proportion of happiness, anger, sadness and joy in their facial expressions during the interview. The AI interview system will give job seekers different star scores.HighestIt's five stars.
Watching the AI interview videos made her laugh and cry. The lighting in some videos was very poor and only the sound could be heard. Some interviewees were recording in their pajamas, and some interviewees would lower their heads to look at the cheat sheets. She could see it all clearly in the camera.
"Unless there is a particular shortage of people for the position, HR will definitely not spend time getting to know the interviewees who score less than two stars or have a bad attitude," said Li Mang. In addition, usually, the final result of the AI interview is based on a comprehensive judgment. The company where she interned would conduct a round of telephone screening before the AI interview, and after the AI interview, a comprehensive assessment would be made by manually reviewing the video.
In Li Mang's opinion, AI interviews have obvious advantages. On the one hand, offline interviews are very time-consuming, and AI interviews allow HR to focus on batch review and synchronize with the business department. Especially for mature HR, they can save time to focus on more difficult positions.
On the other hand, AI interviews are suitable for low-threshold, high-competition positions. The former include sales staff, service staff, couriers, etc., while the latter are high-competition positions for fresh graduates. The workplace skills of fresh graduates are actually similar. AI interviews can screen out some people based on logical ability, expression ability, etc. Some positions have high requirements for English proficiency, and AI interviews can also perform efficient testing.
However, AI interviewers are not as interactive as real interviewers. Some candidates seem to be camera-phobic and appear particularly nervous, uneasy and anxious in the video. Li Mang said that she also watched her own AI interview video in the background, which was "weird, like a robot, with dull eyes."
From the development of AI interviews, there are both popularity and controversy. In addition to discussions on AI bias, fairness and other issues, many job seekers are now demanding more information exchange between AI interviews and applicants. They hope to have more right to know, receive a copy of the results and feedback of AI interviews, and make the results of AI interviews more convincing.