AI Job Interviews: a Capitalist's Dream or an Applicant's Nightmare?

The rise of AI is a hot topic that has captured the attention and imagination of the world, seemingly infiltrating almost all aspects of our daily lives, including now that of our professional prospects.

Source: Daniel McCullough (Unsplash)

It was 4 a.m. when Tina finished her very first AI interview. She was automatically sent the link right after applying for an internship. For those 15 minutes, she lost a full night’s sleep preparing since 10 p.m. While giving everyone a chance, the company spent no time on anyone.

AI assessment and interview provider, HireVue, is making this capitalist dream come true. Trusted by over 888 companies worldwide, especially global giants such as Unilever, Hilton, and Mercedes-Benz who have a huge number of openings and applicants, HireVue decides the futures for millions. Branching out of IT, HireVue is now expanding into finance, education, and hospitality. For instance, it has cut Hilton’s recruitment process from six weeks to five days.

Surely, understanding humans through codes is a great leap in efficiency. On the screen, candidates see their video, a timer, and questions. 500,000 data points are extracted by scanning videos of them answering five questions chosen by the company, each with 30 seconds to think and three minutes to answer. From there, algorithms assess the scanned videos for content, tone, and emotions.

It’s a one-man play for a picky invisible judge. ‘As a computer science student, I know what a machine likes: another machine.’ Tina chooses 4 a.m. at the quiet library for a reason. To be AI-readable, she needs clear pronunciation at a steady pace, no background noise, and a confident look against a white wall. Rather than being freed from physical constraints, candidates are fitting themselves into these standardizations.

As her friend Henry gets immediately rejected for eyerolls when thinking, Tina fears being too natural and human. The eight-year-old technology is in its infancy. The baby freaks out over new things: birds chirping outside, hands touching the chin. Yet, it’s assessing human competencies. In this one-way interview, candidates cannot ask whether they’ve been understood either. If abled 20-year-olds struggle, let alone anyone else. Between humans and AI, there’s an opaque screen, an illusion of transparency.

Nonetheless, having completed over 30 million interviews, HireVue believes it’s ‘democratising hiring’ proven by science. Unilever’s former Chief Human Resource Officer Leena Nair tells the Washington Post that with HireVue ‘the more digital we become, the more human we become.’ According to her, AI improves Unilever’s diversity by reducing lookism biases.

HireVue’s Chief Technology Officer Loren Larsen notes that AI follows rigorous criteria and training. Revealing that ‘interview assessments are up to 29% facial expression, depending on the role’s selected competencies’, HireVue opens a Pandora’s box. But apparently, we prefer them hidden in the ‘black box’ of human thinking. If so, AI has become the scapegoat for all the wrongs of hiring.

Tina is shocked by such comments. ‘When AI is still in dire need of being fed human-generated sources, overrating it is problematic for both parties. Soon, people will start assuming that HireVue is AI at its best. As an aspiring developer, I think that’s an offensive representation.’

The ultimate problem is whether people need to be good at talking to themselves in front of a camera. Unless we’re working with AI, we don’t yet need an army of automatons. Humanity isn’t meant to be machine-like. The real appeal of HireVue is the time it saves for companies. For candidates, even if it’s just as flawed as humans, it adds another layer of discomfort. With HireVue, the giants are sending a terrifying message: we will give everyone a chance even though we don’t really mean it – false hope is mercy.

Imagine spending six hours putting on your best clothes for a 15-minute date with a mirror. Remember, it’s dating 200 others too.

Special thanks to Xinyu (Tina) Hou and Julia Smith for their contribution to the article.