Zhang Liangji of AI Recruitment, and the wall ladder of human applicants
Editor’s note: This article is from WeChat public account “ Brain Body” (ID: unity007), author of Tibetan fox.
The performance peaks of the social animals are not dedicated to all kinds of shabu-shabu, but they are used to deal with the interviewer.
However, it’s not so easy to mix the interviews with “Li Ren”. Since the artificial intelligence fire, many companies have also been handed over to the AI in the hands of the AI, so that the “10-level” job search path has been caught in the back of the fate.
Since last year, many school calendars have also added AI interview/Digital Interview to the interview. Many recruitment platforms have used the intelligent interview system as a business growth point and played the role of a technology fire-transmitter.
And for the applicant, knowing that you are going to face the AI is not tired, the feelings, but also the “roadblock” of the autumn, I am afraid many people will feel Alexander. Especially for the fresh graduates, in the past, there were also the words and deeds of the seniors and sisters who could face the new thing of “AI interviewer”. There is really not much effective experience in the market for reference.
Let’s talk today, where is the border of the AI interview, and how to “crowd” it.
After the AI flood, the employer’s conspiracy
If you see that your favorite company or position needs to pass the AI first, don’t ask for it. Strategic contempt and tactical attention are often the preconditions for the AI interviewer under the “Raiders”.
The reason for this is that there are too many “golden fathers” who use the name of AI and promote the public. Among them, the fast-moving consumer goods sector, which is closest to the mass-scale market, is the hardest hit. Most of the first “AI interview” signs in the school recruiting and other links, such as food and beverage (Coca-Cola), daily chemical products (Procter & Gamble, Unilever) and so on.
On the one hand, most of these corporate positions have no special professional restrictions, and they are wearing a transnational 500-strong aura. The pressure on resume screening is very high. At the same time, the school recruiting is also a very good “boss attention” among young people. Timing, many companies will play the slogan of “finding the future manager”, the competition is extremely fierce, and the creative recruitment methods such as Open Question, AI interview, game evaluation and so on can also effectively help them expand their influence and strengthen their brand image.
This also determines that its AI interview system needs to meet two core appeals:
First, the timing of launch is not fast enough, so that it can be “first to win” in publicity; second, the diversity of data and the robustness of the algorithm are not high enough to avoid detonation due to discrimination, pass rate, etc. The negative emotions of the candidate.
Therefore, the AI interview solutions used by these brands are mostly based on the experience of third-party algorithm companies combined with some of the more mature AI technology applications on the market.
This means that the AI interview system can only play a certain auxiliary role in the initial rough screening. It is difficult to fundamentally decide whether or not the offer can be obtained. Instead, the interviewer can be based on emotional judgment (such as Yan). The bias caused by the value, accent, graduate school, etc., allows more candidates to match the opportunity;
On the other hand, the training logic based on the deep neural network model, as well as the current technical ceilings such as NLP, face recognition, and emotion algorithms, also gives candidates the opportunity to break through the maps of the AI system.
Let’s take a look at these special interviewers with different abilities.
Strategy difficulty one star: Q&A AI interview
This product is often combined with NLP natural language understanding technology and voice recognition algorithms to obtain key information related to job matching in a question and answer format.
The Japanese talent company En Japan has let the graduating college students face up to one mobile phone that only asks questions.An hourly, one-question interview with 126 questions. Seeing the job seeker to the last words that have been abused are speechless…
This kind of “interviewer” uses sound algorithms, statistical speech, intonation, etc., to analyze the response speed and psychological emotion of the candidate. In addition, the NLP algorithm is used to analyze the keywords and semantics of the answers. Issues related to business operations and job requirements, such as the “P&G Eight Questions” and its variants that must be asked by FMCG, to make an initial judgment on the matching degree.
In the face of these interviewers who only choose people based on their voice and language characteristics, because their parameters are artificially set, some implicit emotional preferences are removed in advance, so they will be relatively fair in judgment. Can also be targeted.
First of all, we know that most of these interviewing softwares are standard questions for setting questions and answers in advance. For example, the En Japan test software mentioned above is a collection of interview questions and answers for the past 15 years, mainly related to basic information, work skills, personality traits and so on.
Sometimes with the human interviewer sometimes talk about the family’s tempering atmosphere, or put pressure on the problem, the AI interview system often only one-on-one interaction and questioning, as long as you do more homework before the interview, understand the employer The corporate culture and the concept of employing people are fast-paced, creative, or global. If you adjust and consciously train your own voice, you can avoid many unexpected situations.
At the same time, just like the teacher before the college entrance examination repeatedly reminds that “the word is beautiful and can be more impressions”, it is best to vote for the AI question, try to use some machine logic to organize the language. Because AI is mainly based on real-time voice transfer, keyword extraction matching, and semantic understanding to determine whether the job seeker wants to match the job description.
Thus, to figure out some of the necessary hard-score hard skills, and to target some keywords in the presentation, such as leadership, national projects, conversion rates, team awareness, etc., more helpful to AI Match and filter.
As long as you follow the map, you may feel that AI is easier than on-site interviews.
Strategy difficulty two stars: video AI interview
If the company you are looking at is more powerful or more technical, then the chances of hitting an advanced interviewer like Video AI will be even greater.
In short, the video AI interview is based on the intelligent question and answer. The AI also analyzes the candidate’s facial expressions and muscle movements in real time to judge the applicant’s answer authenticity, personality tendency, and multi-dimensional examination candidates. Whether people are close to the ideal candidate for business.
After listening to the hearts of many of my classmates, I’m afraid that I’m half-finished. Isn’t it possible to “sell” myself with the expression language like flipping my eyes and looking around, now I’m going to watch “lie to me” (micro-expression psychology as the theme) Can you save the American drama?
If you want to recruit such an interviewer who knows how to look at it, in addition to the necessary network check and grooming (see the human interviewer, you should also pay attention to it), I am afraid I have to prepare from the technical cognition.
It can be assured that the use of AI to achieve facial expression emotion recognition, the algorithm does not have sufficient scientific basis, even the AI giants such as Microsoft, Google, IBM, its emotion recognition algorithm is not rigorous, application It is very likely to be seriously misleading in the recruitment scenario. Therefore, the practice of job search software to monitor emotional filtering of job seekers is also considered inappropriate.
For example, an occasional frown does not mean “anger”, nor does it mean that the interviewer must have the qualities of being difficult to cooperate. The place where video interviews work more is actually identifying the best performers.
For example, HireVue or Sonru used by large groups such as Goldman Sachs, JPMorgan Chase, KPMG, Unilever, L’Oreal, etc., the principle is to identify 15,000 features, including selected language, vocabulary of use , eye performance, sound size, etc., then combine these traits, based on the characteristics of the previous “successful” candidate database, to judge a person’s reaction, emotion and cognitive ability.
Finally, by ranking algorithm, a certain percentage of the best candidates will enter the next round.
It is understood that the Hilton Group has used its algorithm to interview more than 43,000 positions. On a global scale, the HireVue system provides 1 million interviews and more than 150,000 pre-employment assessments every quarter.
And as its technical director said, “The extreme complexity of human language, body language and expression requires very careful attention to algorithmic biases and potentially harmful effects,” if the client puts 90% of job seekers on some topics. If they are all screened out, it means that the scope of the inspection point is too small and will change.
After understanding the basic principles of this video AI interview, you will find that you don’t need to be careful with AI. We can certainly find some small tips that are good for the algorithm, but it is recommended that you play as usual, because every habit may affect the pleasure of your future work.
For example, only 10%-30% of the score weights are determined by facial expressions, and most of the rest depends on the interviewer’s speech performance. In the vocabulary, use some characteristic words that meet the preferences of the target company. Job seekers like to say passive words or active words, often use “I” or “we”, whether to use technical vocabulary frequently, etc., will affect the system’s assessment of matching.
Again, for example, the tone of the voice, if some people are really slow, may not be suitable for work such as telephone consultation, and if it is too fast, the user does not have enough time to understand. Using empathy to find the most suitable state for your favorite post, perhaps “feeling” is more reliable than data.
Troubleshooting difficulty Samsung: Reading AI
Have you heard the bleakness of the social animals? Don’t worry about sorrow. If you are “unfortunate” to interview the AI system that reads your social network, you can only buy a lottery ticket to comfort yourself after “streaking” under the algorithm.
At this level, AI tends to analyze the daily behavior of job seekers based on some complex data, and then derive their match with the position. Not long ago, California startup Predictim used NLP technology and computer vision technology to scan the history of Facebook, Instagram and Twitter for job-seekers, and to predict whether they might bully or harass others, whether they might have a bad attitude towards children, etc. Wait.
Of course, such an interviewer was quickly boycotted by the industry. Facebook believes the company has violated a ban on developers using this information to review job seekers, thus greatly limiting the company’s access to user data on Facebook and Instagram. Twitter also interrupted Predictim’s access to its API on the grounds that it was prohibited from using Twitter data for monitoring purposes.
Similar algorithmic risk has also occurred on LinkedIn, the job site, because third-party website HiQ collected LinkedIn data to predict when employees might leave.
The reason why technology giants are clearly dividing the boundaries of such AI interview systems is that machine learning cannot reliably explain the nuances of intonation and speech, such as irony or jokes, applied in interview scenarios. It is very unstable; at the same time, such algorithms are not supervised, that is, they have black boxing, and some employees who may be dedicated to work are likely to lose their jobs without knowing the reason or being able to explain.
In addition, the decisive factor in the success of the interview is given to the AI, which obviously does not conform to technical ethics. If a company directly uses small-scale, singular data sets for training because of policy makers’ preferences/bias, the fairness of AI interviews will disappear, and it may even exacerbate discrimination and exclusion of companies in terms of age and ethnicity. problem. As Anna Cox, a professor of human-computer interaction at the University of California, Los Angeles (UCL), said, “Any data set will be biased, which will exclude those who are really good at it.”
At present, it is still a controversial future for AI to analyze the complex elements of the recruitment interview process.
However, the wheels of technology are accelerating. IBM announced that it would use Watson to actively search the internal training system for data on employee training and learning to determine if they have potential for promotion.In order to conduct an internal assessment. And from such relatively structured data to the interview recruitment, perhaps also to the majority of job seekers.
It can be said that in the past we have seen some areas of mechanization and data structure, and the instruments, translation, and recognition have been replaced by AI. Today, the field of interviews that are full of emotional sensibility is not spared. Fortunately, the magic of technology is fading, and the human beings who hold the sword of knowledge will eventually find the best way to “work with” AI. Iterate and optimize it in the contest and confrontation again and again, and finally get out of the pain of human-machine running-in.