How to get hiring right in the age of AI

If you handle hiring, generic AI-generated cover letters are probably a familiar foe by now. Nearly two-thirds of job seekers are using AI to help craft their applications.

It’s understandable. In a world where some job seekers are having to send up to 50 applications to land a role, tools like ChatGPT enable them to cast their net wide and increase their chances. 

But this spray-and-pray approach to job hunting is a headache for hiring managers. It’s driving the volume of applications up and the quality down, making it harder to spot great candidates.

The natural knee-jerk reaction from HR is to start playing a game of “I spy AI.” If we can just root out the automated applications, we can keep it fair and find the genuine players, right?

The problem is that this approach can give employers a false sense of security. Hiring teams assume that deploying AI detection tools means they’ve solved the problem, and it stops them from digging deeper. 

Robust AI detection tools have a role to play in certain situations; and the tech that powers them is rapidly improving. But they should be just one tool, not the only tool, in hiring managers’ toolkits.

If we want to hire the best humans, we need a deeper fix. We need to evolve our hiring processes, and this starts by removing the elements of the hiring process that AI can easily automate.

CVs and cover letters are the worst offenders, and should have been scrapped long before the advent of ChatGPT. Research shows that the information they present, like names, pronouns, and career gaps, tells us very little about a candidate’s aptitude or skill. What they can do is trigger unconscious bias around what “good” looks like. AI CV screeners carry the same risks: Unless trained on ethical datasets, they can perpetuate historical inequalities.

One solution is to switch up this process, introducing new ways to screen and assess candidates by objectively testing for role-relevant skills. 

A skills-based hiring process, which uses skills tests such as work samples and cognitive ability assessments, demands deeper engagement from candidates. This means that unless specifically designed to evaluate AI skills (which they can be), they tend to be harder for AI to “game”. 

They’re also far better at predicting an applicant’s future performance than proxies on CVs, and can help tackle the application volume problem: The extra engagement that skills tests require is the antithesis of the “spray and pray” approach. It acts as a filter, with only those who feel invested in the role going on to apply. 

An honest, crystal-clear employer brand does something similar. It enables employers to attract fewer, but better-suited candidates. So, employers should ask themselves: Is it currently clear to candidates what you represent, how the team is structured and what benefits you offer? Do job seekers know whether your company is office based, hybrid, or remote? Being transparent on your company website, social media, and job adverts about the whole package—including salary expectations—can help narrow your candidate pool to applicants who want what the company offers.

HR managers should also consider how else they can leverage AI to their advantage. There are plenty of ways it can support hiring teams beyond detecting candidates’ AI usage; for example, to help with accurate candidate scoring, automate interview scheduling, and analyse data to predict future job performance. Just be careful that the models you’re using, particularly to screen candidates, aren’t trained on a singular image of success.

Above all, it’s important to remember that even if we can catch AI-wielding applicants, are we sure using AI is a skill we want to penalise? I’m not advocating for generic “AI slop” applications. But since approximately 30% of work activities could be automated as early as 2030, taking a leaf out of Anthropic’s book and testing applicants’ AI literacy as part of the hiring process for relevant roles is savvy. 

We may not be able to eliminate poor-quality AI applications altogether, but acting like detection is the sole solution to our broken hiring processes is a false economy. It risks blinding us to the deeper work that needs to be done to get hiring right. By establishing stronger, more targeted pipelines and robust assessment practices, hiring teams will be able to find and attract top talent with the skills needed to thrive in the modern world of work.

View the full article