AI in recruitment: A guide for employers on policies and fair evaluation

The jobs, they are a-calling!

Recruitment is a full-time effort for growing companies as they seek to fill their open positions with talented individuals wh-o can contribute their time, energy and innovative ideas that fuel organisational growth. Artificial intelligence (AI) is transforming the world of recruitment by giving recruiters new, cutting-edge capabilities that help them hire the best candidates for open roles. From spotting cheating at interviews to new-age tools for better onboarding, AI is indeed a game changer.

There is an important aspect of recruitment that has been silently pervasive and difficult to root out to date: bias. Bias in recruitment, whether favourable or unfavourable, can mean that the wrong candidate lands up in a position at best. At worst, this insidious trait can hurt organisational growth in the long run.

Yet, bias is inescapable. All of us accumulate biases with every encounter we have with the world at large: biases about situations, people, backgrounds, cities, languages and cultures. We favour something in every instance, or we avoid something. This is because we are all too human.

This is where AI can be useful. Technologists and recruiters have particularly worked together to eliminate human bias during that all-important step of screening. Let’s explore how.

AI-powered resume screening software operates on predetermined criteria, such as specific skills, candidate diversity criteria or certifications, using keywords. Such screening is particularly useful when sifting through thousands or tens of thousands of resumes that land in inboxes from corporate career pages, job boards and college recruitment drives. It is manually impossible for a limited number of recruiters to sift through such numbers without letting biases in, such as recency bias. Even more pervasive are biases such as the similarity bias (where the candidate(s) exhibit some affinity to the recruiter), the gender bias or the halo effect (when some positive trait of the candidate overshadows everything else).

Consider statements such as:

• “The candidate is from XYZ College. He must be a good candidate!”
• “She is in her early twenties. Can she stick on in this job for a few years?”
• “There are so many candidates! Let me just pick the last twenty I reviewed.”

This can happen in the best of organisations, and AI can mitigate this in one fell swoop. Indeed, automated candidate screening is no longer just a fringe benefit: it is a mainstream feature that can turn into a competitive advantage for many companies!

In a nutshell, AI-powered screening can help the enterprise with value-based recruitment, a strategy that puts the right candidates in front of recruiters and hiring managers. Value-based recruitment is beneficial to the recruiting organisation in many ways: besides burnishing the employer brand, it also contributes to a peaceful environment at work and ensures that more engaged employees are at work, resulting in higher productivity.

What, then, about bias in the AI itself? The lack of transparency within the AI algorithm being deployed, and any resultant biases it brings into the process, has emerged as a significant challenge to hiring bodies. Here are some specific bias mitigation strategies HR leaders must employ:

1. Ensure your algorithms train on diverse and representative data.

   AI is only as good as its training data. If the data is skewed in any manner, the resultant algorithm will exhibit bias. HR leaders must tackle this from training data onward.

2. Perform regular audits and checks of the AI.

   AI is an adaptive technology that learns as it operates on new data. Regular checks on the algorithm and any necessary corrections are imperative to ensure no biases creep in over time.

3. Implement transparency in algorithms.

   AI cannot operate as a black box. The candidates whose data AI operates on must be given information on how it operates and what the logic is behind certain decisions. Otherwise, trust gets eroded and this can hurt the brand, besides opening up the possibility that the best candidates may not apply for the vacancies.

4. Provide adequate data privacy and security safeguards.

   It bears repetition—AI is only as good as its data. The data must be safeguarded. Sensitive candidate information must be handled confidentially. HR leaders must communicate the scope and purpose of any data collection to candidates and seek permission if necessary.

Not surprisingly, these very strategies form the bedrock of a fair recruitment and evaluation process that is enabled by AI. Crafting an effective recruitment strategy powered by AI from screening onwards takes time and effort. As discussed above, it is rooted in the pillars of representative data, security, transparency, audits and checks.

When done right, AI-based screening can allow your recruitment to fly, giving you speedy access to the best available talent for your open positions! Talk to HirePro, the new-age AI-powered recruitment platform, to understand our AI-based screening tools. As an ISO 27001-certified and GDPR-compliant organisation, HirePro’s technology is built around the pillars of fair technology-powered recruitment. Let’s make your hiring Fearless!