In its first, yet intriguing opening scene of the popular TV show ‘Lie to Me’, the main character heroically saves a church from blowing up, by reading a prisoner’s body language and micro-expressions to find the truth about the bomb’s location.
Check out the scene here:
The main character of this TV-show is based on Dr. Paul Ekman, a psychologist and expert on body language and facial expressions at the University of California. Dr Erkman explains that it takes sophisticated technology and years of training in order to master the skills needed to read such micro expressions. However, these micro expressions do not seem to be tied to culture or ethnicity but seem to be rooted in biology. A skill that might also be very useful in a job interview.
With the vast advances AI technology over the last years, it has become more common place for computer programs to be able to recognize pictures, voices, objects and predict behavior based on it. Just to give some examples, think of Facebook’s AI that asks you to tag your friends in pictures or Amazon’s Alexa. Alongside these advances much behavioral analysis has been done on various available data sets. A Stanford research group for instance, managed to judge personality traits better than friends with as little as little data as 10 Facebook likes (The article can be found here). Even the sexuality of people can be predicted with a high accuracy purely based on pictures taken from the internet (Link). It might be alarming to some that a computer program can predict who we are so accurately. However, especially in the recruitment space such technology could save a lot of time and money.
Exactly this is the space that the innovative, US based, startup HireVue has entered. HireVue provides a digital interviewing system that uses video interviews to assess the ability of candidates. Behind the system – a powerful AI that tracks your eye movements, the level of stress in your voice, the language you use and in total over 15.000 traits to identify top performers. This promising technology has already been adapted by many large firms including J.P Morgan and Goldman Sachs. When applying to these companies it is no longer sufficient to send your resume and a letter of motivation. These firms now also ask you to log into the HireVue platform and record yourself while answering a set of questions. This video material is then analysed and based on HireVues technology and the top performing candidates are invited for a second round of interviews. While this might at first seem ‘freaky’ officials at J.P Morgan and Goldman Sachs have pointed out clear advantages of using this technology in their hiring process. First, using the software is much more scalable and allows these banks to interview many more people. Second, by eliminating the human beings from the interviewing process any form of human bias is taken out of the hiring process which could be great for avoiding discrimination.
But how does an HireVue interview look like in practice?
In the case of Goldman Sachs applicants can expect 5 questions for which each they will get 30 seconds to prepare themselves and 2 minutes to speak to the camera. The software is very strict with the time limits and does not give any second chances. If everything goes well the interview is over within 15 minutes and candidates can expect to hear back within 2 weeks. (If you are curious what kind of questions can be expected at these check out this article.
In conclusion, in the future we can expect AI to impact many areas of our day to day life, including the hiring process. It might, at first seem that programs such as HireVue decrease discrimination and allow for more diversity in the hiring process. However, we also have to keep in mind that these AIs learn from the data that we chose to input. In my opinion, this leads to two problems. First, if ‘top performer traits’ are always only assessed based on historical data, a company might be slow to adapt itself to changing ‘top performer traits’. Second, within 15.000 different personality traits it will be increasingly difficult to understand all the different correlations. This could also lead to some form of discrimination, this time however, it is more difficult to find someone who is accountable.
Although I find this new way of interviewing potential candidates pretty amazing, I must say it will make the whole application process event more stressful than it already is. Knowing that there is a machine that monitors your movements, your mimics and even detects and analyses the nuances is your voice is kind of scaring for the applicants. The introduction of AI and ML in recruitment process transforms the interview into a type of interrogation, which surely influences applicants performance. Additionally, by removing the human interaction from the process will prevent recruiter from using their gut feeling when making the selection decision. Even if this new technique, will potentially eliminate the effect of human biases, it will also prevent the managers from using their intuition, that is based on their experience in working with people. I also agree with the problems you have presented above. Thus, in my opinion, before implementing AI into their recruiting process, companies need to ask themselves: are machines smart and experienced enough to judge people’s character, skills and abilities better than we as humans are?
First of all, thank you for your reply and your good points! To address your comment V. maybe I did not explicitly mention this but the process is not meant to completely replace a human interview but can be seen as an addition. In the current reality, as Marcin has pointed out, most recruiters only spent about 6 seconds on each CV which will often lead to ‘false negatives’ due to the interviewers bias. This means that the managers will reject a candidate not because he would not be able to fill the position but just because they do not have enough time to properly look at the entire CV. By including an additional layer of interviews to the process and by making the information from the video interview available to the managers they might be able to make a better recruitment decision. What do you think of this technology if we do not completely eliminate human interactions? I am looking forward to hear your thoughts!
Thanks Laurin for your interesting post about the role of artificial intelligence in improving the recruitment process. I have to admit that I haven’t heard about the HireVue before, but it seems to be an amazingly valuable tool, especially for the initial screening of candidates.
As an addition to the solution you shared with us, I would also mention the HR platform called Pomato. According to the research conducted by Ladders, recruiters spend on average only 6 seconds reviewing a candidate’s resume. To my mind, it is not enough to objectively assess potential skills of a candidate. Moreover, selection of candidates may also be biased due to some personal recruiter’s preferences, thus making the whole process ineffective.
The Pomato Platform has been developed by IT managers and performs over 200,000 computations on candidates’ resumes within just a couple of seconds. IT leverages machine learning algorithms in order to assess their skills and rank them in an appropriate order. In the end, the recruiter sees a dashboard with strengths and weaknesses of all potential candidates. Pomato’s main goal is to support recruiters, who are lacking an adequate IT background in order to properly assess IT talents. Nevertheless, I am certain that we will soon see similar solutions for the other industries as well.
Whenever I read about all these great innovations, however I cannot imagine that we can completely exclude a human interaction in the recruitment process. In the end, we are all social creatures and very often a manager who recruits us, will probably become our direct boss. His goal is not only to look for the candidate who can be a top performer, but also to find the person who suits his team. In the end, Since we are not all the same, some subjective factor is also needed to make a proper decision. We have different preferences and we like to work with different kinds of people. The question is whether an AI algorithm can understand this puzzle as well?
Sources:
https://cdn.theladders.net/static/images/basicSite/pdfs/TheLadders-EyeTracking-StudyC2.pdf
https://www.tlnt.com/how-ai-is-improving-recruiting-and-hiring/
http://www.pomato.com/
http://recruitingtools.com/artificial-intelligence-recruiting-tools/
Hi Marcin, thank you very much for this interesting extension to the blog post! I really like the additional information that you have provided especially because I think that this software is already in use for many industries. I just did some research on this and it seems that many large companies already do a first automated scan of your CV. Some research by Deloitte even suggest that as much as 60% of all CVs undergo a first automated scan (http://blog.bersin.com/are-applicant-tracking-systems-now-a-commodity/).
Also I do agree that the profile between different departments does change a lot and that many things cannot be determined by this AI. However, I do believe that such technology will make it much easier to do a first selection which is then followed by further interviews with the departments. I am definitely curious what the future holds in this regard and I am sure to start writing my CV in a way that a machine can read it as well thanks to your input! If you are interested what you can do in order to write your CV in such a way check out this link: https://www.simonstapleton.com/wordpress/2015/04/02/how-to-create-a-machine-readable-resume-in-12-steps-and-why-you-should-do-this/
As a future job applicant it is great to learn about the latest updates in job interviews! In my opinion, this system can definitely ease the recruitment process of companies but at the same time would make it even more unpleasant and stressful for the candidates. It is usually the case that during job interviews applicants are very nervous and if they are aware that there is an AI system tracking their every single move or mimic, they can become even more stressed and perform worse than they can do otherwise. Also, sometimes the candidates fail to understand a question correctly and if there is no interviewer they can ask for clarification, they can totally mess up the interview. I also believe that having interviews the old way, with a recruiter, is a good opportunity for the company to test a candidate’s social interaction skills and also to ensure that both the candidate and the company get to know each other well. If we only count on a machine asking the questions, the communication is only one way and the applicant won’t be given a chance to ask anything they want.
Hi there! Thank you for your great comment! I agree that some people might become even more nervous if they know that their every movement is tracked by this technology. However, doing this automated video interview also has the advantage that you can do it from where ever you feel most comfortable and that it will be very straight forward questions. Additionally, do you think that a persons social skill can be determined by focusing on someones voice, body language or micro expressions and that it could therefore be read by a machine? Also to clarify, I the candidates will still have another interview if they are successful in their first ‘AI round’ where they will be able to get ask questions about the company.
To extend this discussion, I think Marcin made a very interesting point about machines reading your CV. Have you prepared your CV in a way that a machine can read it? If so, maybe you can share your experience and give us some tips and tricks about how to best do it?