Posted by 
Recruiter and HR managers’ challenges are more significant than ever, and the stakes are higher. The Great Resignation and the current labor shortage in the US, are but a few of these challenges. However, being a recruiter or HR manager is also a great time. There is an increasing amount of powerful tools to help them meet these challenges, notably in the advancements made in AI and machine learning.
SwissCognitive Guest Blogger: Gergo Vari, CEO of Lensa – “AI and Big Data Reducing Recruitment Biases”
We’ll begin with a few definitions and explanations, then move on to highlight ways to take advantage of the advancements in AI as well as things to be on the lookout for.
What Is Recruitment Bias: examples and definitions
A bias is a sort of prejudice: an illogical distortion of expectations based upon preconceived notions. Bias can work against someone or in their favor. It is an inclination or “feeling” someone has. It is unwarranted and unfair, and as it pertains to recruitment, it can lead to less-than-optimal hiring results. In some cases, it can even be illegal and can leave the employer vulnerable to lawsuits and fines.
Conscious vs Unconscious Bias
A bias is not necessarily intentional. In fact, in most cases, the person with a bias is unaware they even have a bias. This is referred to as an unconscious bias.
For example, a recruiter might have had a negative experience with a person who attended a certain university (let’s say, for example, the University of Minnesota).
When this recruiter comes upon an application from a candidate who also attended the University of Minnesota, the recruiter immediately has a negative feeling about the candidate. And they might not even know why that is. This negative bias is likely to influence how the recruiter views the rest of the candidate’s application. It might influence how they read the candidate’s resume and cause them to pass on what might otherwise be an ideal candidate.
A conscious bias works much in the same way as an unconscious bias, with the notable difference being that the biased person is aware of their bias. It is easier to deal with and mitigate the effects of a conscious bias; knowing is half the battle, as it were. However, this form of bias can take on highly nefarious forms such as sexism, racism, classism, and other nasty isms.
Potential Consequences of Recruitment Bias
More often than not, it is a company’s employees that drive the business and are ultimately responsible for its success or failure. Recruiters and HR managers are thus tasked with the weighty challenge of staffing the company with the best possible employees, making sure those employees stay motivated, engaged, and productive, and then retaining those successful employees.
Recruitment bias has the effect of limiting the talent pool from which recruiters draw talent. A smaller talent pool means a smaller chance of successful recruitment.
Furthermore, recruitment bias alters the criteria from which talent is selected. Where there might be a correlation between a candidate’s background and skillset to the likelihood of their success with a company, recruitment bias introduces other variables that are unrelated to a candidate’s chances of success. It’s akin to playing blackjack with a faulty understanding of arithmetic. You’re unlikely to have much success if you think three is four and jacks are worth five, for example.
How AI and Big Data Are Used in Recruitment
Artificial intelligence (AI) analyzes millions of data points taken from sources such as job offers, resumes, performance reviews, sales reports, etc. The data points come from a variety of sources, such as the company’s internal records, and industry statistics. These data points are collectively referred to as big data.
From a constant analysis of big data, AI systems are able to detect patterns, identify correlations, and make predictions or quantify the likelihood of desired and undesired outcomes.
AI is used in all phases of the recruitment process. In many instances, AI is used to carry out repetitive or menial tasks thus freeing up the recruiter to do more important things. It is also used to further engage with candidates and thus increase the chances of a positive candidate experience. With so many uses, it’s no wonder AI and automation are enabling the future of business.
However, for our purposes, we are going to focus solely on the uses of AI that help to mitigate (or eliminate) recruitment bias.
Identifying and Defining the Candidate Profile
From an analysis of big data, AI systems can create the ideal candidate profile, one which yields the greatest likelihood of a successful employee.
- How important is the candidate’s academic background versus their prior work experience
- What skills should be required
- What kind of work experience transfers best for the position needed to be filled
More importantly, AI systems can identify criteria, patterns, and correlations that wouldn’t be intuitive or immediately apparent to recruiters otherwise.
Filtering Out Candidates
is not only through an analysis of the candidate’s resume that AI can determine the likelihood of a candidate becoming a successful hire. AI systems also conduct skill assessments and performance tests which generate more precise data points that allow AI systems to make more informed decisions as to which candidates should go on to the next phase in the recruitment process and which candidates should be dismissed.
Refining the Recruitment Process
AI systems are free of many forms of bias just as they are free of ego or pride. This means that AI systems can conduct regular analyses of the recruitment process itself and its own role in the process. Based on a recruit’s acceptance rate, retention rate, and performance evaluations, to name but a few of the relevant data points, AI systems work to continuously refine and improve the recruitment process – free of bias with only the quantifiable desired outcome as its target.
Eliminating Recruitment Bias: A Considerable Benefit of AI
Biases, in all their forms: conscious, unconscious, a bias for or a bias against, are an integral and inescapable part of the human condition. We try to recognize them and minimize them, but they form a part of who we are. On the other hand, AI systems are not susceptible to biases. They operate exclusively on quantifiable, fact-based results (to the extent of their programming).
Unlike humans, AI systems do not have “bad days.” They are not susceptible to mood swings that could affect their performance or skew their decision-making process.
AI systems make decisions or recommendations based solely on the algorithms they are programmed to respect. This entails certain limitations, of course. But when used in conjunction with human decision-makers, it constitutes a big step toward result-based decisions and, by definition, a significant reduction in both conscious and unconscious biases (if not an outright elimination of these biases).
Why It Is Important to Reduce Recruitment Bias
We’ve already looked at a few of the negative consequences of recruitment bias: a reduced talent pool from which to draw talent means a reduced chance of successful recruiting, potential lawsuits and fines, and potentially missing out on the best talent for your company.
Now, let’s take a look at some of the benefits a company is likely to enjoy once they’ve reduced or eliminated recruitment bias.
A Reduction in Recruitment Bias Translates to an Increase in Diversity in the Workplace
Diversity in the workplace comes in many forms: employees from different backgrounds, different ethnicities, different skill sets, different perspectives, and outlooks.
With diversity in the workplace, a company enjoys many benefits and advantages that a homogeneous workforce cannot provide.
- Flexibility – A diverse workforce greatly increases a company’s ability to change and adapt to new circumstances and challenges.
- Innovation – A diverse workforce also means a workforce that approaches processes and problems in a variety of ways. Examining things from different perspectives is part and parcel of innovation. It is how new things are created, new improvements are implemented, and new solutions are found.
- Customer Loyalty – Today’s consumer, seeks more than a good product or service at a good price. Today’s consumer seeks to assert their identity through their buying habits and through their association or patronage of brands. For this reason, it is crucial that a company’s workforce resemble its customer base. This further increases the chances that the customer will be able to and willing to identify with the company or brand.
A Reduction in Recruitment Bias Translates to an Increase in Employee Retention Rate
An employee is far more likely to stay at a job they are successful in, one in which they are able to express their talents and add value. It is in both the employee’s and the employer’s best interest that the employee truly be a good fit for the position and responsibilities they are taking on. When this is the case, the employee is far more likely to stay engaged and productive and is far less likely to leave prematurely. Reducing recruitment bias increases the chances of a successful employee-position fit and thus increases a company’s overall employee retention rate.
Potential Drawbacks: What To Be Wary Of
An AI system functions solely based on the directives (or algorithms) it is given. This is the strength of AI, but it is, in some ways, its weakness, too.
When it comes to reducing recruitment bias, it is quite possible that a particular algorithm be corrupted with a bias. Since an AI system is making decisions, suggestions, and predictions based on data points that were accumulated with inherent biases, if attention is not paid to rectify these errors, these biases will only be perpetuated.
For example, an AI recruitment system will invariably analyze resumes collected over a period of years. What happens when the vast majority of these resumes come from male candidates or if the collection of resumes fails to contain examples of minority groups? In these cases, the AI system may be susceptible to perpetuating this imbalance.
The concern has caused many industry insiders to mobilize against this potential drawback.
For example, legislators in New York City have recently passed an AI Bias Law, which is slated to go into effect on January 1, 2023.
This law is an expression of concerns many industry experts have on the limitations of AI in recruitment. And while the New York law is the first of its kind, it will certainly not be the last. this is only one example of trends in AI that we should keep an eye on going forward.
Suggested remedies for AI bias include conducting regular bias audits. These audits are meant to screen the AI’s algorithms to look for any potential biases based on race, gender, and ethnicity. These audits should be conducted by an independent or third-party auditor and never by the creator or designer of the AI system.
The New York AI bias law also mandates that the results of these audits be made available and displayed on the website of the company using the audited AI system.
In a Nutshell
Recruiters and HR managers are tasked with the crucial responsibility of staffing a company with the best possible employees. However, as recruiters and HR managers are human, they are susceptible to conscious and unconscious biases, which have a detrimental effect on the recruitment process. Using AI systems, machine learning, and big data can greatly reduce the likelihood of the recruitment process being contaminated with bias. However, vigilance is needed to ensure that the algorithms the AI systems rely on are not, themselves, contaminated with bias.
About the Author:
Gergo Vari is Lensa CEO. He has one mission: to revolutionize job search for companies and professionals. His journey through founding, funding, and exiting successful startups has taught him a valuable lesson: the hiring process is broken. He shares the need for recruiting and human resources technology that puts people first
Der Beitrag AI and Big Data Reducing Recruitment Biases erschien zuerst auf SwissCognitive, World-Leading AI Network.
Source: SwissCognitive
