Uncovering and Addressing Bias in HR with Machine Learning
Despite our greatest efforts, it’s difficult for humans to avoid bias. Whether conscious or unconscious, the biased decisions we make can affect the people around us. When companies—and more specifically, human resources (HR) departments—are exposed to bias, it can affect numerous business decisions such as hiring, promotions, and opportunities for employees. This ultimately can affect employee satisfaction and retention within the company. So, what can we do to identify the biases that we don’t even know we have? One promising method is machine learning.
By using machine learning (ML) and artificial intelligence (AI), HR departments can screen for inherent bias; predict employee churn and work to retain those identified as valuable; and find channels optimal for recruiting the best talent. In addition, ML can be used on the development side to tailor courses to better train individuals, as we’ve already seen in some schools.
Sounds great, right? But even with all these benefits, companies are still hesitant to implement machine learning in HR. They don’t want to take the ‘human’ out of ‘human resources’ just yet.
Game-Changing Predictive Machine Learning Radio Series
On the latest episode of the Game-Changing Predictive Machine Learning radio series, host Bonnie D. Graham drilled down into the implications of human bias in HR and how machine learning can be used to address it. (Listen to the replay).
As always, she was joined by experts in the field who shared their own thoughts and insights about the topic. This week’s guests were:
- Jeff Mills, Director of Solution Management for Talent Acquisition at SAP SuccessFactors
- Erin Roberts, Professional Services Consultant at SAP SuccessFactors
- John Schitka, Solution Marketing Manager at SAP
The Panel Talks: Why Should We Implement Machine Learning in HR
So, what was the panelists’ take on why it has taken so long to implement machine learning in HR if it has been shown to be so beneficial?
“It’s always hard to let go of control, but to let control go to a ‘non-human’ is a very difficult thing,” sympathized John Schitka. “You have to realize that the machine isn’t making the decision, it is making a suggestion. Humans still make the decisions and actions. But, machines can do a very good job at showing some of the biases we have and making us aware of that bias, so that we can counter it.”
John explains that people shouldn’t fear that machines might make decisions on their own that might not align with the company’s intentions. The goal of implementing machine learning and AI in HR is to simply provide unbiased suggestions based on the data it’s provided.
HR workers can then use these suggestions going forward. For example, if an HR worker chooses to hire a new employee on the basis of their unconscious bias, the AI can screen for that bias. It will then warn the HR member about their biased decision and suggest an alternate candidate who would be better suited.
Erin Roberts added to this thought, saying that we’ll likely never be able to solely rely on machines for making decisions, especially in HR. After all, a machine cannot tell what is going on in an employee’s personal life, but an HR professional might know based on having a friendly conversation with the employee.
Therefore, there is a need for both machine and human interaction when making decisions. Ultimately, machine learning is only as effective as the data it’s provided with—and this is where some concern does come into play.
What You Give Is What You Get—It All Comes Down to Data
If a machine’s suggestions are only as good as the data provided to them, then what kind of suggestions will the AI make if it receives biased/inaccurate data?
Simply put, if a machine receives data based on bias, then it will continuously make decisions based on that bias—and there is no value to that. So Jeff Mills stresses that it’s not just about giving the machine lots of data. It’s about giving machines the ‘right’ data so that the AI can make unbiased, proper suggestions.
In addition to this, businesses need to leverage and incorporate their very broad datasets, enough so that when new data is inputted, the machine is able to identify and omit potential biases, and create suggestions based on fact and hard data instead.
Putting these thoughts together, we can see that there is a great opportunity for addressing bias in HR with machine learning. If a broad-enough set of the ‘right’ data is inputted into the AI, it will be able to uncover our unconscious bias and warn us about it. It can then provide unbiased suggestions that are good for the company, the employees, and for business. But, if the AI receives the ‘wrong’ data which itself contains bias, then there will ultimately be no difference between humans and machines.
Sure, the machine may not be perfect, but neither are people, and that is the point of machine learning. The AI learns the more it is used, and as it adapts, machine learning and AI can become an exceptional aid for HR departments when making critical decisions.
Want to hear more from these three game changers? Listen to the full replay online, and join us next time on November 7th from 2:00-3:00pm ET live on the VoiceAmerica Business Channel.