Subscribe
 Vetting Before Going Virtual

Vetting Before Going Virtual

Managers should neither be blind champions nor Luddite opponents. They must first understand the technology, and thereafter, partner with providers to address key issues.


There has been a lot of discussion over the virtues of interviews conducted by virtual interviewers developed via Artificial Intelligence (AI). Such interviews present advantages such as - the interviews can be conducted at the candidate's convenience, more number of potential candidates can be assessed, and answers to the same question by different candidates can be compared to assure fair selection. Also, such systems can better capture minute observation of facial characteristics, delays in response, and other nuances used for qualitative evaluation of the spoken response. However, as these systems evolve, there are certain concerns that must be addressed and fixed before virtual interviewers get wider acceptance.

 

AI-systems, such as those driving virtual interviews, are often 'trained' by presenting them with a large number of images and sound files of people tagged with descriptions of their likely emotional states. This way, the machine learns to spot if a person is confused, evasive, confident, enthusiastic, tired, or disinterested as the interview proceeds. If this training material features people of only one ethnicity or gender, the algorithm will gain greater efficiency at analysing people of the same ethnicity or gender, but may give erroneous results when processing people from other backgrounds. Companies developing such systems are increasingly being asked to share the specifications used by them to create the training inputs. In many cases where such a disclosure has been made, it appears that the diversity within that set is limited. If training inputs have featured Caucasians alone, it is likely that an African or an Asian may get her responses flagged incorrectly (E.g. 'confused’ when she is may actually be ’attentive'). This suggests an action point for managers evaluating such systems for deployment within their organisations. They must ask the vendor to share details of the training set to judge if the diversity therein is appropriate for the pool of target candidates over which those systems will be used.


There is another point of discomfort in such virtual interaction, which is somewhat harder to address. Nestor is a French company developing AI powered teaching assistants. Their system monitors the facial expressions of a student as she accesses a video lecture. At the end of the lecture, it presents a quiz to test retention and comprehension, and makes sure it  picks questions from those sections of the lecture where the student appeared disinterested or otherwise inattentive. This forces the student to be attentive at all times. However, students in Paris schools, where it was first tried, have not been happy with such intrusive monitoring. It makes them uncomfortable to be watched so minutely all the time. This is true of the virtual interviewing sessions too. One likes to see the person who is asking the questions. If she looks attentive and empathetic, one becomes more enthusiastic in one's response. If one senses incomprehension on the faces of the interviewers, as a candidate one can repeat or elaborate what they have said. Such an option has not been included in virtual sessions. Also, the candidate likes to know the people watching her and to what end. While this is obvious in a physical setting, it is not the case in a virtual setting. The recorded video may be seen by any number of people for a variety of unspecified reasons. The candidate may be uncomfortable with her performance being played over and over again before people she has not met. 


It is important to get these softer features right because just as the hiring company is assessing candidates, the candidates are also assessing the hiring company. In a situation where there are many candidates and limited jobs, hiring companies may feel that they have the upper hand,  and may not care much about the candidate experience during the interview process. That would be a mistake. flood people are scarce at all times, and therefore, normally have more than one option. Even if they accept a position out of necessity, first impressions die hard, and they may be open to alternate positions when they become available. In the extreme case, a bad interview experience may lead to a rejection of the job offer in itself. The case of Olivia Bland is instructive.


Olivia Bland is a young professional who applied for a job with Web Applications UK earlier this year. She was upset with the aggressive interviewing where she was ridiculed, and her work severely criticised by the interviewer, who was the CEO of the company. During all this, there were two other people present in the room, who merely watched her and asked no questions. This added to her confusion and stress. After all this, she was offered the job. She turned it down and took to Twitter to narrate her experience. Her post went viral and was commented upon in many media platforms. One knows many interviewers who take pride in asking irrelevant questions and discounting all responses by the candidate in order to create stress. They believe this helps them assess the candidate's ability to deliver under stress. Olivia may be an exception, but many candidates facing such an unempathetic environment may choose to see the job being offered only as a stop-gap arrangement, till a better one shows up.

A positive interview experience helps in 'selling' the job to a candidate.


She may find it that much harder to turn down an offer after a positive first experience. At the very least, she may consider working with the company sometime in the future, and may also generate good word-of-mouth publicity amongst other potential hires. Technology is forcing change in all processes in every organisation. Facial recognition and analysis is one of the many drivers. One would be foolish to resist change merely because it is uncomfortable. All change is. In fact, Marcel Saucet, the man behind Nestor, believes that since facial analysis is going to be all pervasive, it is best that students get used to it early, no matter how uncomfortable they feel. As he puts it, "We cannot go against the natural laws of evolution." That is certainly an interesting, if otherwise tough, view. In practice, Managers should neither be blind champions nor Luddite opponents. They must first understand the technology, and thereafter, partner with providers to address key issues so that their organisations can ride the coming wave without getting submerged.
 

Gautam Brahma is a management consultant who advises start-ups and SMEs on strategy & operations including sales, HR and IT. He carries an experience of over four decades in the public, private and non-profit sectors in telecommunications and IT industries. He has been an invited speaker on multiple industry forums and a monthly columnist on HR issues for nearly two decades. Gautam is based out of Gurgaon and can be reached at [email protected]

Comment

0/3000 Free Article Left >Subscribe