In the evolutionary arms race between interviewer and interviewee, I think it is inevitable that both roles will at some point be played fully by robots. AI is already helping us to filter through CVs – one day, we will be able to leave chatbots entirely to it: everywhere, in pockets of cyberspace, one large language model will be offering another a seat and asking about the last challenge it faced at work, while we humans get on with something more useful.
We came one step closer to this utopia recently, when one – clearly quite brilliant – job candidate was revealed to be using AI to feed her answers during a Zoom interview. A phone app recorded the questions in real time and delivered “perfect” replies, which she calmly read off the screen, thus demonstrating innovation, resourcefulness, and a healthy disrespect for the whole interview process. I hope she gets the job.
This disrespect is, after all, long overdue. It may be time to get rid of the job interview altogether. Since at least the early 1900s, it has squatted in the centre of the hiring process, where it has revealed – primarily – that we like to think ourselves “good at reading people”, when in fact we really aren’t. We know this because the interview has been the subject of a swathe of research. And what has this research told us? In sum, that if one candidate outperforms another in an informal interview, the chances that they will do better in the job are little better than flipping a coin.
How do we get people so wrong? Well, one major problem is bias . How a candidate looks tends to matter more than it should – the beautiful always do better, even when the job involves data input or working for radio. People also tend to give jobs to those most similar to them in terms of background, gender, age and race (there are now attempts to train recruiters out of this, but biases are hard to shift). When recruiters aim to find someone who is a “cultural fit” for their workplace, this is often what they are doing, consciously or not.
People tend to give jobs to those most similar to them in terms of background, gender, age or race
Then, too, minds are often made up during the first few minutes of an interview, in the bit where you chat about the traffic or the weather, supposedly to get the candidate to relax. This suggests superficial qualities weigh heavily in hiring decisions, whether or not recruiters are aware of it. The firmness of a handshake can be used to predict offers, even when grip strength has little to do with the job itself. Apparently, this is in part because first impressions can dictate the direction of the rest of the interview. If recruiters feel apprehensive about a candidate at first glance, they might be inclined to ask them tougher questions, or look for evidence that their impressions are correct.
In his book Noise, psychologist Daniel Kahneman provides a telling example of this sort of bias. Two colleagues interview the same candidate, who explains that he left his last job because of a “strategic disagreement with the CEO”. But the colleagues interpret this differently. One, who starts with a positive view of the interviewee, takes it “as an indication of integrity and courage”. The other, who has formed the opposite impression, believes that instead it shows “inflexibility, perhaps even immaturity”.
This wouldn’t matter, perhaps, if interviews were treated as a relatively small part of the hiring process – the final flourish. But they tend to leave vivid impressions, which can override CVs, references, and even test scores. Yes – performance in one highly artificial situation seems to matter more than actual data.
Despite all this, employers are deeply attached to the process – they remain convinced that they cannot really “get a feel of a candidate” without it. Like driving or sex, we all seem to have a deeply held belief that we are good at interviewing. Structured interviews – where every candidate is asked the same question and evaluated according to an algorithm, rather than according to the guts of their interviewers – are better at predicting job performance but have been fiercely resisted by employers. They prefer to trust their intuition to tell them whether a candidate is right or not. They “just know”.
One answer, then, as to why the interview remains in the hiring process, is that it massages the egos of recruiters. I think that this might also explain another puzzle – a fad for off-the-wall questions that have nothing to do with the job.
Such questions have long infuriated job seekers. When in 1921 the American inventor Thomas Edison interviewed graduates at his plant, the questions included “Who wrote Home Sweet Home?” and “What is the weight of air in a room 20x30x10?”. “ ‘Victims’ of test say only ‘a walking encyclopedia’ could answer questionnaire” ran a headline in the New York Times. But, of course, the trend didn’t end there. “If you could be remembered for one sentence, what would it be?” Google once asked candidates for an associate account strategist position. Goldman Sachs, meanwhile, had this question for prospective bankers: “If you were shrunk to the size of a pencil and put in a blender, how would you get out?”
Of course, one effect of such questions is to make a profession seem far more interesting than it is – thus flattering the interviewers.
But hiring is too important for this sort of nonsense; a nation’s success, after all, rides on the quality of its employees. Getting the right people into the right jobs is where fairness and productivity meet. We should start by making job interviews more structured. We could end by getting rid of them altogether.