The use of AI and automation in recruitment has grown at speed in recent years. There are hundreds of tools on the market that profess to match candidates with vacancies or undertake reference checks at the press of a button.
But in doing so, AI has worsened an already terrible situation, while also creating a whole new set of challenges, adding insult to injury. This is because AI can’t fix a broken system. It can only accelerate it and find workarounds. And the current recruitment process is beyond repair.
A failed system
Consider for a moment traditional hiring practice. At its heart is the CV, which has barely changed since Leonardo da Vinci first introduced the concept some 400 years ago.
In a bid to deal with such outdated documents and manage the information within them, CVs are saved in centralised databases. These are a hacker’s delight, offering a wealth of personal data for identity theft. Inevitably, there have been high profile cases of data breaches. For example, when the US recruitment site Ladders exposed 13 million user profiles.
Meanwhile, job adverts engage in a ‘spray and pray’ approach. They’re often badly written, poorly targeted and result in either too many unsuitable applicants or not enough relevant ones. This leads to the Herculean task of wading through CVs, 63 per cent of which hold misinformation, costing businesses £23 billion annually.
Then comes the highly subjective interview where people are turned down for ‘not being right’ before the laborious job of referencing successful candidates. Inevitably, a third of those hired leave within 90 days. To say this is a poor process is an understatement of epic proportions.
A troublesome solution
To tackle the situation, practitioners have embarked on one of the biggest mistakes any digital transformation project can make: trying to automate a bad process with AI.
Why on Earth would anyone automate a broken system? It’ll certainly speed things up, but it’s like putting a McLaren engine in a Morris Marina. Start it up and the wheels will soon fall off. Not to mention the doors, seats and windows.
Yet this is exactly what’s happening. And it’s creating a raft – actually, a super tanker – of problems. Firstly, with such a damaged and fragmented framework to operate within, AI’s been forced to make decisions about who gets hired or even who sees job ads.
This can lead to indirect discrimination and bias. In fact, UK GDPR rules restrict employers from making solely automated hiring decisions. The ICO is now investigating rule breaking. The EU and US are also legislating.
Secondly, once AI has chosen candidates, it has the task of referencing. Or in other words, checking to see if a candidate is lying. Really? If the hiring process worked, referencing wouldn’t exist. Every word on a CV would be true.
Rather than address the problem, AI begins digging through hundreds of databases to get every detail it can about the applicant. Credit scores, social media posts, tax status, education, convictions. It’s muckraking on an industrial scale. Perhaps even stalking. There’s no regard for relevance to the job, let alone privacy. In short, employers are overstepping the boundaries massively.
Thirdly, in doing all of this, AI draws upon huge volumes of data that humans can never hope to understand. As a result, it cannot ‘explain’ its decisions simply. The AI becomes a sealed black box that holds people’s futures inexplicably within. It obscures the thousands of turns it took to get to the destination precisely because the journey was so arduous.
With this in mind, HR leaders need to ask themselves whether they want to allow people’s livelihoods to be decided by a supercharged bull rampaging through the china shop of recruitment at a thousand miles an hour. It’s not good for jobseekers, could be in breach of the law and puts the reputation of employers under the spotlight.
