DiversityIT/Tech/DigitalTalent TechnologyImproving AI to Reduce Unconscious Bias in Recruitment

The Use and Concerns of AI  Artificial intelligence (AI) techniques such as machine learning, sentiment analysis, and natural language processing are used in the hiring process to optimize and automate components of the process. Previous research has shown that unconscious bias in hiring can occur in several areas. Candidate sourcing, screening, and outreach are the most frequent recruiting processes for which AI is being used to optimize and automate.  AI’s increasing use in sensitive areas,...
Cinthya Soto4 months ago101212 min

The Use and Concerns of AI 

Artificial intelligence (AI) techniques such as machine learning, sentiment analysis, and natural language processing are used in the hiring process to optimize and automate components of the process. Previous research has shown that unconscious bias in hiring can occur in several areas. Candidate sourcing, screening, and outreach are the most frequent recruiting processes for which AI is being used to optimize and automate. 

AI’s increasing use in sensitive areas, such as recruiting, has sparked a discussion over bias and fairness. It’s long been known that AI-driven systems are prone to their creators’ biases – we unknowingly build biases into systems by training them on biased data or using criteria set by experts with implicit biases. To solve this problem, AI models and scoring algorithms should be constructed to be blind to anything that isn’t related to the job, and then evaluated to ensure that the models are predictive, fair, and don’t produce false results.

AI has the potential to assist humans in making more inclusive decisions in recruitment—but only if they work diligently to ensure that AI systems are also fair. 

Why AI Needs Improvement 

According to an article by Mckinsey & Company, underlying data rather than the algorithm itself are most often the main source of the biased AI. AI is the simulation of human processes by machines. AI is learning from humans entirely, meaning it is the biased data set used to train the algorithm that needs to change. The deepest-rooted source of bias in AI is the human behavior it is imitating. 

In an interview with Forbes, Mike Hurdy, Chief Science Officer at Modern Hire says, bias in AI mainly comes from the data that is used to train the models. Using data that is convenient and readily available rather than data that is highly job-relevant is often where bias starts. For instance, many models are based on resume or job application data. This data is experience-based, which tends to be inherently biased and research shows, not a great predictor of ultimate success. 

Reducing Bias in HR with AI 

To get to the root of the problem, we must first improve AI and then apply these improvements in the recruitment process. 

Improving AI 

During his interview with Forbes, Mike Hurdy also states that helping to ensure candidates are reviewed based on the content of their response and not on other non-job-relevant criteria begins with how the technology is set up. Making sure that AI models and scoring algorithms are created by being blind to anything that is not relevant to the job, and then validated to ensure that the models are predictive, fair and do not create adverse impact, is essential. But, like other tools in the hiring process, AI tools need to be continuously audited to ensure they’re accurate, fair and free of bias. 

For AI to be efficient in recruitment, we must also make sure that companies are using unbiased data to train AI. According to an article from Eightfold.ai, AI requires a broader data collection than any single organization can provide. By looking across massive data sets from a wide range of sources, AI can more easily differentiate the noise of bias from the signal of relevant skills. To improve the hiring process, AI tools need to focus on what matters to success in any given position, as well.

Additionally, Eightfold.ai states that AI-enabled hiring software has improved as our understanding of bias has changed. Today, cutting-edge AI can more effectively combat unconscious bias with broader data sets and a focus on skills. 

How can AI reduce unconscious bias in recruitment?  

According to an article by Harvard Business Review, AI holds the greatest promise for eliminating bias in hiring for two primary reasons:

​​1. AI can eliminate unconscious human bias. 

Many current AI recruiting tools have problems, but they can be fixed. The beauty of AI is that we can program it to fit specific requirements. One fundamental notion is that AI should be built in such a way that it can be audited and any biases are removed. 

2. AI can assess the entire pipeline of candidates, rather than forcing time-constrained humans to implement biased processes to shrink the pipeline from the start. 

We can only eliminate bias by utilizing a fully automated top-of-funnel approach that shrinks the initial pipeline to the capability of the manual recruiter.

The Future of Using AI to Reduce Bias in HR 

It is impossible to correct human bias, but it is demonstrably possible to identify and correct bias in AI. If we take critical steps to address the concerns that are being raised, we can truly harness technology to diversify the workplace.

Human control is still required to guarantee that AI does not replicate or introduce new biases depending on the data we provide. If AI uncovers a bias in your hiring process, you now have the chance to correct it. We can use our human judgment and knowledge, aided by AI, to determine how to correct any biases and enhance recruitment procedures.

So, while technology can detect the problem, it is still up to recruiters to come up with solutions to overcome biases that may be restricting workplace diversity.

Cinthya Soto

Leave a Reply

Your email address will not be published. Required fields are marked *

Comment Policy | Privacy Policy

Related Posts

{{ image }}

{{ title }}

{{ date }} {{ comments }}
{{ readingtime }} {{ viewcount }}
{{ author }}
{{ image }}

{{ title }}

{{ date }} {{ comments }}
{{ readingtime }} {{ viewcount }}
{{ author }}
{{ image }}

{{ title }}

{{ date }} {{ comments }}
{{ readingtime }} {{ viewcount }}
{{ author }}