Skip to main content

Hello. It looks like you’re using an ad blocker that may prevent our website from working properly. To receive the best experience possible, please make sure any ad blockers are switched off, or add https://experience.tinypass.com to your trusted sites, and refresh the page.

If you have any questions or need help you can email us.

AI job interviews are good for only one thing

Job applications are written using AI and are scanned by another AI and then, if you get through to the interview stage, you’ll be interviewed by AI too. Surely this can’t be right

Job applications are written by AI, scanned by AI and increasingly conducted by AI too. Image: TNW/Getty

Job hunting has never been especially enjoyable. But it’s become outright odd. That’s because companies and candidates have succumbed to the siren song of AI. Recruiters and HR departments use LLMs to write their job descriptions, and candidates use LLMs to adjust their CV to suit a specific application. They also use it to write their cover letter, which is in turn scanned by another piece of software on the employer’s side. And round and round we go, until the planet runs out of water.

On top of this strange AI dance of machines basically talking to each other, you may now be subjected to an AI interview. And I mean “subjected to”. Let me run you through it. 

I’ve done a few interviews with “AI recruiters” and they followed the same script. First, you receive an automated email from the company you’ve applied to, inviting you to complete your AI interview. There is no workaround offered. If you want to go to the next stage of the recruitment process, you have to click on the link provided and register on their platform, or the third party platform of their choosing.

There, you are told that to proceed with the interview, you must agree to be recorded and give the website access to your computer’s microphone, camera, and more often than not, authorise it to access your full screen. The latter is to make sure that you do not use AI whilst being questioned by an AI, because that would be cheating.

The interface looks like a classic video call except in place of your interlocutor’s face, there is a blue shiny sphere that glows, very much a corporate interpretation of the Eye of Sauron. At least, you don’t feel guilty if you spend the whole interview looking at your own camera feed.

Depending on the position you apply for, you might just need to “talk to the machine’’ or you will be given timed tasks to do as well (writing, editing code, translating…)

Unsurprisingly, the AI software I encountered had been given women’s voices. Maybe as a dark homage to the women who used to work in HR. Soft yet robotic, the pacing of these voices is always uncanny, because AI doesn’t pause to breathe.

The most common types of AI-led interviews are asynchronous interviews, where question prompts have been fed to the AI and the interviewee gets one to three minutes to give an answer; and conversational AI interviews, where the software tries to react and build on your previous answers.

I did both. In the first case, it led to some very awkward moments. The software kept lagging, leaving extremely long pauses between my answers and the next question. Being a human who is used to filling silences, after a bit, I would start talking again, but so would the software. And then both me and it would stop. And so on, and so on.

Another time, during an interview supposedly made to test my French, I was asked the same question three times in a row. “Could you describe your hometown, in your native language?” There is just so much one can say about a rural village in Brittany. It’s green, there are cows, and bad wifi.

Conversational AI interviews are not much better. To make it look like a conversation, the AI often just repeats your previous answers, and then changes the subject completely. Also, I heard a lot of “that’s wonderful” or “that’s impressive”, as reactions to some of my answers that were neither of those. It was very reminiscent of ChatGPT’s sycophantic tendencies.

By far the hardest part of having to do AI interviews is not knowing how or by whom they will be reviewed. Or if they will be reviewed at all. Nor are you being told clearly where the videos of the interviews are stored and for how long. Or if they’re safe.

The one consolation – that if you did very poorly at least nobody would see it – is not even guaranteed.

Indeed, in March of this year, Mercor, one of the largest online recruitment platforms for AI companies, was hacked, compromising the personal information of its 40,000 contractors, including the video interviews. Despite the data risks, an increasing number of companies are adopting this technology as a standard screening process.

I remember reading an essay by David Sedaris where, after a particularly humiliating event, he tells himself “Someday, this will be funny”. And I heard myself thinking that while doing those AI interviews. They are so bad, they would be hilarious, if it weren’t for the fact that, for many people, jobs are very hard to come by at the moment.

My personal view is that this gimmick, sold to businesses to make them look “AI ready” and to save a few quid, will not stick around too long. I’ll personally start a movement to boycott AI interviews if I have to.

Perhaps it should start with the employers themselves. Perhaps they’ll eventually realise these AI interviews are only good for one thing: indicating to their possible future employees what they really think of them. That they’re disposable. 

Hello. It looks like you’re using an ad blocker that may prevent our website from working properly. To receive the best experience possible, please make sure any ad blockers are switched off, or add https://experience.tinypass.com to your trusted sites, and refresh the page.

If you have any questions or need help you can email us.