Is AI Resume-Builders Can Remove Potential Recruitment Bias?

Bias is a pressing issue when it comes to recruitment. Artificial Intelligence (AI) applies lessons learned from millions of datasets to address bias-prone aspects of a resume. In a detailed discussion with Michelle Armer, Chief People Officer at CareerBuilder, we look at how AI resume-builders are tackling this challenge. We also share key takeaways for HR managers to remember when exploring AI-assisted recruitment.

Research has consistently shown that resumes can significantly impact hiring decisions, even when it is not justified. From a 2012 Yale study, which showed that professors were more willing to hire and generously pay male fellows, to a recent report highlighting the impact of ethnic-sounding names on hiring, it is evident that biasis a very real and prevalent issue. Often, this can be prevented by carefully articulating one’s resume, so that recruiters can focus on qualifications and achievements without allowing any bias to creep in.

We spoke with Michelle Armer, Chief People Officer at CareerBuilder – a platform that applies artificial intelligence (AI) to help create job-optimized resumes – to understand how its AI resume-builder uses advanced algorithms to detect and prevent any bias-prone areas in resumes.

How Do AI Resume-Builders Work?

Unintended bias could enter a candidate’s resume due to lack of familiarity with the English language, personal tonality preferences, or inadequate understanding of resume writing best practices. Unfortunately, when recruiters look at these resumes during high-volume hiring cycles, they might prematurely reject a candidate without delving a little further. And this affects both sides: the organization loses out on top talent while the candidate has to look at other, possibly inferior alternatives.

AI resume-builders can solve this by applying insights learned from previous hiring trends, data gathered from resumes, and recruiter preferences. They can spot and eliminate grammar issues, improving chances for candidates whose primary language isn’t English.

“CareerBuilder’s AI resume-builder helps remove bias from the job search process from the beginning. The tool ensures tone neutrality, fixes grammar mistakes and tightens language. This can help individuals whose primary language is not English, as well as those who have little downtime to perfect their resumes, to be recognized for their skills rather than overlooked for the job due to errors,” adds Armer.

Interestingly, the AI resume-builder can also translate skills learned from one field into the potential for another. For example, veterans can turn their military experience into civilian job titles to ease their transition into the job market. This can help recruiters find skills-based talent in an economy where the labor pool is limited and the demand for talent is at its peak.

How Does an AI Resume-Builder and Recommendation Engine “Learn”?

Any AI resume-builder software must be built on a bedrock of clean and comprehensive datasets. Furthermore, this should be continually updated so that the AI engine can pick up on the latest trends and make the most appropriate suggestions. Machine Learning (a subset of AI) helps to automatically gather and refine insights without requiring human intervention. And by removing the need for human intervention, the possibility of bias is also reduced.

Data sources that the AI engine can work on include public job boards, candidate profiles, resumes that have been processed via text analytics, titles and skills that are commonly searched, as well as candidate screening data.

“Over 200 data scientists are utilizing datasets that have been collected, parsed and normalized from over 2.3 million job postings, 680 million unique profiles, 310 million unique resumes, 10 million job titles, 1.3 billion skills, and 2.5 million background checks,” Armer revealed, discussing how CareerBuilder provides data-driven insights to candidates.

Such tools can also form the foundation of AI-powered recruitment engines. Companies can use similar tools based on similar datasets to ensure bias-free hiring in their organization.

Key Takeaways for Organizations Looking to Adopt AI-assisted Recruitment

While AI can remove bias potential from resumes, human-developed AI engines also risk replicating the inherent prejudices of its makers. Michelle calls this the difference between more technology and better technology: “When employers rely solely on more technology vs. smart technology, that’s when hiring managers fall victim to situations like their resume sorting tools favoring men over women for certain positions, or overlooking qualified candidates because of semantics.”

In their bid to hire fast in a competitive labor market, hiring managers should not indiscriminately adopt AI-based solutions. “Technology plays a pivotal role in the Hello To HireTM process – the entire candidate experience where employers have to plan, find, screen and hire the right talent, but it is not the final solution,” believes Michelle.

In an astonishing experiment conducted by the University of Virginia, it was revealed that image datasets of activities like cooking, shopping, and sports were gender-biased. In fact, AI algorithms mistakenly labelled a man cooking as “woman.”

Hiring managers must keep such risks in mind when choosing technology. Only then can the brightest candidates, represented by optimized resumes, be aligned to their most fitting and deserving job role.