A week ago, I had applied for a Junior Data Scientist position with a large conglomerate based in the UAE. They informed me that I was shortlisted for the first round, and sent me a link for an online assessment. The assessment was conducted by a third party called HireVue, and it consisted of video interviews, and some gamified tasks. I came to know that these tests would be evaluated by an AI algorithm, and not by a human. Needless to say, I was curious.
A day after taking the test, HireVue sent me their insights from my interview. Although I agree with most of their findings, some are up for debate. They did get a lot of things right about me from just an hour of interview, and testing, which is impressive. I found the whole experience interesting, and I thought I would write a short post about it.
What is HireVue?
According to their website, HireVue is a pre-employment testing, and video interviewing platform. They offer solutions for pre-hire assessments, video interviewing, game-based assessments, coding assignments, and scheduling interviews. Their clientele includes impressive names, such as, Singapore Airlines, Vodafone, Rackspace, etc.
What do they test for?
I am going to quote verbatim from from HireVue:
“We offer two types of automated assessments: game challenges and interview questions. Games assess how you interact with and process information. Interview questions assess how you communicate, the content of your message, the meaning in your message, and the emotion you express. Once we have your responses, we use an algorithm to identify behavioural patterns tied to the competencies that lead to success in this role. Insights associated with our algorithms have been scientifically validated.”
How do they test?
There were two parts to the interview process: a video interview with five questions, and a series of games that tests psychometric abilities. The video interview starts with a single question. You get 30 seconds to prepare a response. You can then record your response as a 3-minute video presentation. Once this is done, you move on to the next question. You continue this way until all questions are answered.
The psychometric evaluation consisted of four sets of games. The games test your mathematical abilities, logical, and reasoning skills through tests, such as, matching figures, and shapes, and so on. Each game had multiple levels, and each level had time limits. The levels increase in difficulty as you go along. Your mission is to do complete as many levels as possible without compromising on accuracy. It has been a while since I played games like this, but I did find it fun.
Tips for your HireVue interview
- Go through the “Tips for your Digital Interview” document first. They would have sent you the document along with your interview invite. It would tell you what to expect, how the assessment will be done, etc. It also includes useful tips on preparing for the interview.
- Scour YouTube for interview tips. There are plenty of resources there on how to prepare for the interview. You can also find common interview questions for HireView.
- There are practice interview questions on HireVue. Make use of this feature to test your presentation skills. These responses are not shared with the client, so use them to get as much practice as you need. These tests are also a good way to test the audio and video quality of your device, so make sure it is crystal clear when you play back your recording.
- The interview and tests took about an hour for me. It is important that you are totally prepared, and free from distractions during this time.
AI vs. Human Recruiters
Companies have been using automation for years to weed through thousands of resumes they get. So, automation tools are not new to the recruiting industry. But, having an AI agent perform the first level screening felt like science-fiction for me. However, with all the advancements we see in Natural Language Processing (NLP), and Image Recognition, is it all that surprising?
Is it better to let an AI do the initial screening? I came across two Wall Street Journal articles on HireVuew:
- How new data-science tools are determining who gets hired.
- A face-scanning algorithm increasingly decides whether you deserve the job
These articles explain how automation is taking over traditional roles, and raises some important points on the pitfalls of using AI for recruitment. As I see it, there are both merits and demerits to using AI. I have put together some points to argue from both the sides:
| AI Recruiter | Human Recruiter |
|---|---|
| Uses metrics from top performing employees as a bench mark to assess future employees. | Recruiters can use their experience to offer a well rounded opinion. They can identify candidates who doesnt coform to norms. |
| Looks at many inputs, such as tone of voice, micro expressions, etc., to assess candidates. | Humans are pretty good at reading cues, but this ability might vary from person to person. |
| The algorithms used by AI systems are proprietary. We may never know the exact parameters under which they operate. It is essentially a blackbox. | One can always ask the recruiter their reasons to accept, or reject a candidate. |
| There might be inherent bias in the training data, which might affect the chances of someone who does not match the system’s expectations. | Humans have biases, and prejudices as well. |
| Can produce consistent results, as long as the training data is consistent. | Humans may not be as cosistent as the AI. |
| Highly scalable. Can cope up with increased demand. | There is only so much work a human one can do in a day. |
| Takes considerably less time to complete the evaluation. Less turnaround time means more people can be interviewed. | Turn around time is more. |
Final thoughts
The more I think about it, the more I am beginning to lean towards the AI system. An AI system can go through a lot of candidates to produce an actionable result for the recruitment team. This makes the initial screening more efficient, and faster. The candidate can then be invited for a face to face interview, and can be quizzed again on other topics of interest, that was not covered during the AI interview, such as, technical skills. This could, in theory, give an over all better hiring experience to the interviewee.