Rawpixel - Fotolia
Why reducing hiring bias isn't easy
Because people may harbor hidden biases, diversity and inclusion initiatives can be difficult to execute. Take an in-depth look at the issue and learn whether AI can help.
Hiring bias -- often unconscious -- is one very pervasive issue that gets in the way of diversity and inclusion initiatives.
As the gatekeepers to employment, HR teams must recognize their biases. Recruiters form both conscious and unconscious biases when seeking out new candidates and may miss out on hiring someone who would excel within the company. Vendors promise AI technology and software can help fix this hiring bias, but it may not always help solve the problem.
In this Q&A, Stacia Sherman Garr, co-founder and head analyst at RedThread Research, discusses her thoughts on diversity and inclusion obstacles, why hiring biases exist and whether AI can fix this issue.
Can you define diversity and inclusion?
Stacia Sherman Garr: Diversity is a variation in backgrounds, beliefs and experiences, with respect to gender, race, ethnicity, language and mental abilities. In a simplified version, there is visible diversity [such as gender], which we tend to see as being legally protected; and then invisible diversity [such as sexual orientation and class], which tends to be more of those things that are not immediately obvious but still influences people's perspectives.
Inclusion is about allowing the equitable and fair distribution of resources within an organization, which allow all employees to be appreciated for their unique contributions and to feel they belong to the formal as well as informal networks within it.
What gets in the way of companies reaching this goal?
Stacia Sherman GarrCo-founder and head analyst, RedThread Research
Garr: One of the biggest things is that, fundamentally, this is about the culture that an organization has, and changing a culture is difficult. When you think about it from that perspective, it means that diversity and inclusion can't just be an on-the-side initiative. It's not enough to just have an employee resource group; you actually need to be reinforcing a diverse, inclusive way of thinking. This factors into how HR teams acquire talent, promote employees, give feedback and coach people, so it is systemic change and that's why it is hard.
Why is there a sudden demand for diversity and inclusion?
Garr: Demographics, particularly in the United States, have been changing. We see an age demographic shift and an ethnicity demographic shift. Younger people, or people from underrepresented groups, are more likely to bring up their perspectives. They're more likely to push issues and are less afraid to do it in a work context than the previous generations.
The second reason is that overall work has globalized, so we're seeing workplaces becoming more multicultural with more freelancers or virtual work.
Third, there is a relationship between diversity and inclusion and business outcomes. There's research that shows this, so organizations are seeing the connection between diversity and inclusion and financial goals.
Finally, #MeToo brought this all to a head. It wasn't just that #MeToo was about sexual harassment. What #MeToo underscored for HR leaders, in particular, was the role of culture with regard to people being treated fairly and equitably in our workplace. If you look at the numbers of underrepresented groups in leadership, you see we've been working on this problem for years but things haven't shifted. The combination of the heightened focus and frustration, plus the technological advances [such as in AI] are why we've seen a lot of the technology come to the fore.
What are your thoughts on the hiring bias as part of diversity and inclusion?
Garr: People aren't necessarily conscious of bias, which makes this issue pretty complex. It could certainly be that every human has a bias toward people who look and talk and think like they do and that can seep into the hiring processes, whether it's the recruiter or the hiring manager. It can be conscious in that they may think, "I have seen X type of person from Y university succeed here, therefore I think that that's what we need in this role," where that may not be the case in terms of those being the necessary factors to result in success.
When we then translate that into this conversation about AI, it's important to note that AI is just advanced math. All it's doing is pattern recognition and learning from previous patterns. If we as humans have had a challenge with bias in the past, then a technology whose sole job is to look at our patterns of the past and deduce from that is going to extrapolate some level of bias, if it goes unchecked.
Software vendors market products that would "fix" this hiring bias. What are your thoughts on that?
Garr: I do not believe it's possible to completely eradicate all biases. I think that there are ways to reduce the biases that exist. With additional analysis capability we can see some of the things that humans have done that have bias within them, or indicated bias on a systemic level, and address those. It should be the obligation of vendors to work on this, and they should be very transparent about what they're doing to address it. But I am skeptical of any vendor that says it has wholly eliminated bias.
In 2018, Amazon had to scrap an AI recruiting tool that showed bias against women. What does that say about the use of AI to improve diversity?
Garr: What it tells us first and foremost is that we shouldn't allow engineers to run rampant without HR intervention. In that instance, HR was actually not at all a part of that technology development; it was done by a bunch of engineers in the business. It also underscores the importance of oversight and testing. Once developers have built something, it needs to go through rigorous tests to understand, "Is there bias here, and if there is, how can we address it?" Technology shows that there's been an event in the past and we need to have some way of foreseeing that for the future. Unfortunately, it's been painted as such, but I do not think it should be used to cast all AI in a negative light.
How important is cognitive diversity today in relation to this hiring bias?
Garr: It is a way for us to bring in different perspectives, to push for new ideas and to build things that haven't been there before. Much innovation comes from the intersection of more than two existing knowledge bases that people haven't combined before, and that is fundamentally cognitive diversity. But it's important to not forget other diversity. There's actually been some studies that show just by having visibly diverse individuals, it actually forces the other people in the group to take a different perspective. So cognitive diversity is important, but we shouldn't forget visible diversity as well.
Let's say companies have hired these diverse employees; what comes next in terms of inclusion?
Garr: Making sure that the organization has a culture that's open to diverse perspectives. There's a number of organizations that are using organizational network analysis to understand how to connect people into the organizations that works more effectively. And then there's all sorts of tools that are available. Historical diversity tools such as employee resource groups or play action committees can help with some of this.
Then there is taking a hard look at all the various talent, practices and processes, and adjusting the organization's approach so that they are open and aware of what's necessary from an inclusion perspective. Heightening people's awareness through all the different practices of an organization of what they need to do to be inclusive is really important.