Originally published in The Times, by Sathnam Sanghera

Last weekend I drove the new Hyundai Kona EV, one of the many electric cars hitting our roads, and it was impressive in all sorts of ways beyond that it doesn’t run on petrol. The emergency braking system detects a potential collision with a vehicle or pedestrian and automatically stops the car if necessary. The lane keep assist system monitors lines on the road and will steer the car if you veer off course. People talk about the self-driving revolution as if it is decades away when it almost with us already.

Would I, however, want this car to appraise my next job application? I ask not to be deliberately whimsical, but because the technology that is driving the autonomous car revolution, and is already being used to help to diagnose disease, develop unmanned police stations in China and play Paul Young over your smart speaker in the kitchen because you liked Softly Whispering I Love You on Spotify last week, is already at work in HR departments.

According to CNBC, “nearly all” Fortune 500 companies are using some kind of automation to enhance hiring processes. The Evening Standard asserted last week: “Firms such as Goldman Sachs and Unilever are using artificial intelligence software to weed out candidates” for graduate training programmes. Meanwhile, a search on the internet reveals a plethora of AI outfits offering HR support, including the software specialists Triplebyte (“Stop wasting time on bad candidates. Get introduced to better engineers today”) and Hirevue (“Predictive assessments tied to higher quality hires”).

How does this tech work? Well, it varies with each company and every client they sign up, but in general a job candidate will sit in front of a webcam or camera on their phone and answer a bunch of pre-recorded questions that pop up on screen. The video of them replying will then be analysed by algorithms for things such as body language, concision, confidence, “micro-expressions” (such as how much they blink or smile or scowl), specific words or phrases and other characteristics that employers may have decided are desirable.

Needless to say, the people behind the tech are bullish about its potential. Among the claims they make is that AI will widen the pool of applicants companies recruit from (given that tech can analyse many more applications than human beings); will improve diversity and remove unconscious bias (robots are not sexist or racist but objective); could eradicate the need for candidates to do multiple interviews with multiple employers (Triplebyte claims: “don’t apply for a job, just do an interview and we will match you to the best job”); and will save companies millions (stripping away the need for HR teams to sift through thousands of initial applications).

This last claim is particularly exciting: it can only be a good thing that there are fewer HR people on the planet. At the same time, I can’t help but feel that we’ve been here before with optimistic predictions of what tech might be able to do for the world, and with Western democracy crashing around our ears as a result of Facebook and Twitter, nations around the world being deprived of vital taxation because of Google and Amazon and tech giants everywhere altering our lives and minds in ways that are disturbing and unpredictable, it might pay to be sceptical.

So, a reservation: AI technology is still work in progress. I mean I love my Alexa and Google Assistants more than some relatives, but they are far from perfect (last week Alexa confused playing a Melvyn Bragg podcast with playing All About That Bass by Meghan Trainor). The Kona EV was great, but some of the driver assistant systems stopped working when it rained and the radar got blocked, and at work a colleague tweeted that our new American expenses system, which scans receipts and uses AI to fill in forms, set the currency for dinner at a Chinese restaurant in Brighton as yuan. I would be nervous about this still-developing tech deciding the direction of my career.

Another reservation: far from making recruitment fairer, you could argue that AI interviews only really measure your ability to perform with a robot rather than do a job (in the same way that conventional job interviews only really measure your ability to develop rapport rather than do a job), and there are signs that people who can afford it are already finding ways of gaming AI for their benefit. For £7,350 (ex VAT), Finito, a City fintech, will offer you everything from networking support to interview guidance, including training to handle AI interviews, until you get a job. The people behind it tell me that a bursary is available for jobseekers needing help with fees, but, let’s face it, it’s another example of how the privileged always find a way to replicate their privilege.

Which brings us to the biggest problem: AI can be deliberately programmed to be unfair. In recent years Facebook has been lambasted over revelations that its technology allowed landlords to discriminate on the basis of race and employers to discriminate on the basis of age. Now a group of jobseekers is suing the tech behemoth for helping employers to exclude female candidates from recruiting campaigns and The New York Times reported: “The employers appear to have used Facebook’s targeting technology to exclude women from the users who received their advertisements, which highlighted openings for jobs like truck driver and window installer.”

I suspect that debating the effects of AI will be the controversy of our generation and that it is too early to come out entirely for or against. The jury is still out and even tech moguls disagree: Elon Musk said that AI was “a fundamental risk to the existence of human civilisation”; Mark Zuckerberg claims that it will “deliver many improvements in the quality of our lives”.

Even at this stage, though, it is obvious that we need to approach the technology with caution and to wait and think before we hand the controls to the robots.

Sathnam Sanghera is a journalist and author. Follow him on Twitter @Sathnam