While artificial intelligence clearly has a key role to play in the future of recruitment, it is critical to ensure that this new technology eliminates hiring bias rather than entrenches it.
With jobseekers applying for more jobs than ever before, it's not surprising that employers are turning to technology to help them handle the load. Back in the days when jobseekers mailed in resumes by the post, they'd only apply for half a dozen positions. The advent of job boards saw this rise to 35, but with all of today's technology it's leapt to an overwhelming 105 applications per jobseeker.
Separating the wheat from the chaff sounds like a perfect job for AI, a technology which is great at sifting vast amounts of data in search of hidden insights. The problem is that AI can't always be trusted to make important decisions which still require the human touch.
Expr3ss! highlights the most promising candidates for recruiters to consider interviewing for that particular job, but it never automatically discards those applicants who don't seem to make the grade.
Many AI systems are often trained on previous decisions, such as letting it study years of hiring data to decide what the perfect candidate looks like.
That sounds like a great idea, unless of course you've been biased for years – unconscious or otherwise – when deciding who to hire. If that's the case, all you're doing is teaching a machine to make the same biased decisions.
Case in point, Amazon was forced to abandon its experimental AI recruiting tool after discovering it discriminated against women. The technology was rating candidates for software developer jobs and other technical posts, but not in a gender-neutral way.
The problem was that this AI was trained by looking for patterns in the resumes of Amazon's successful applicants over a 10-year period. Of course, this only entrenched the existing male dominance across the tech industry: because men usually got the job in the past, AI learned that men must usually be the best candidate for the job.
The system even rewarded candidates for masculine language in their CV, while downranking anyone who went to a women's college.
At least Amazon recognised what was happening and put a stop to it. Some organisations wouldn't even see the problem, because the AI's bias would align with their own bias, so they wouldn't notice any bias and would be convinced the technology was doing a fantastic job.
Even "blind hiring" techniques don't fix the problem of bias, as studies have found that AI can pick up on gender signals in a CV, even when the name, pronouns and other obvious giveaways are removed.
Thankfully, lawmakers are starting to crack down on the problem. In New York, employers which use "automated employment decision tools" must tell candidates. These employers must also submit to annual independent audits to prove their systems are not sexist or racist.
Expr3ss! predictive hiring is very different, because it doesn't skim through resumes looking for telltale signs of what it assumes a successful candidate should look like.
Instead, Expr3ss! uses short surveys to assess every applicant on their merits. It looks at skills, attitudes and temperament to determine which applicants would be a great fit for your team, and then recommends who should make it through to the interview stage. It's only a recommendation, no one is cast aside and hiring managers can still ever applicant.
This approach is going to become more important as both businesses and jobseekers use AI in an attempt to game the system.
Another key recruitment trend is that some businesses are using AI to write job advertisements. Meanwhile, some applicants are turning to generative AI tools like ChatGPT to write their resume and cover letter, in an attempt to bluff their way into an interview.
Related article: ChatGPT ghostwrites CVs
At this point, businesses will be using AI to write job advertisements and then using AI recruitment tools to evaluate resumes. But many of those resumes will be bogus, written by generative AI that's trained to read job advertisements and then write resumes that tell recruitment AI tools exactly what they want to hear.
It gets even worse when employers start using AI-powered chatbots to interview candidates, who simply cut and paste their answers from ChatGPT.
In this ludicrous situation, everyone is flying blind – you've got biased AI making bad hiring decisions, based on bogus resumes and made-up answers written by dodgy AI. What hope has such a business got of hiring the right person?
The result is a string of bad hiring decisions, which creates a revolving door of employees and takes a massive toll on the business. Meanwhile, the best candidates are constantly overlooked because they don't seem to fit the mould.
There's no question that AI will play an important role in the future of recruitment. The real question is whether businesses will blindly trust AI to do the job, and pay the price, or whether they'll use AI wisely to assist human recruiters in making the best hiring decisions.
