CXO

How machine learning can help companies eliminate bias in hiring

At SAP's Sapphire Now conference in Orlando, SAP product leads and machine learning experts discussed how diversity can be woven into hiring strategies.

On Tuesday at the 2017 SAP Sapphire Now conference, a panel of top product experts at SAP and AI researchers discussed a critical issue facing businesses: Avoiding hiring bias.

The panel included Brenda Reid, product management at SAP SuccessFactors, Patricia Fletcher, solution management at SAP SuccessFactors, Anka Wittenberg, chief diversity and inclusion officer at SAP, and Yvonne Baur, product management at SAP SuccessFactors, and was moderated by Gabriela Burlacu, principal human capital management (HCM) researcher at SAP SuccessFactors.

When it comes to diversifying the workplace, many people say "We get the 'why,'" said Reid. "But where's the 'how'?"

Other panelists emphasized this point, illustrating why it's essential for businesses to embrace diversity initiatives. "We're starting to see people say 'How do I do this?'" said Burlacu. "Not just in HR, but across the business."

SAP SuccessFactors HCM Suite, a cloud-based solution, is an effort to help businesses begin to develop a diverse workforce and eliminate bias in hiring and management decisions. The goal, according to a press release, is to "embed bias prevention, detection and elimination into how [businesses] work and make decisions." That's especially important when you think about a company like SAP, which has a global workforce of 80,000 people.

Burlacu illustrated some of SAP SuccessFactors' tools that can prevent bias in making management decisions, like photo-less calibration—since photos, it has been shown, can lead to unconscious bias when making judgments about employees. Also, a "gender indicators" tool can pinpoint potential bias at a company by highlighting how the different genders are rated by management.

20170516145747-3.jpg
Image: Hope Reese/TechRepublic

"Chief diversity officers need to be in the room where important decisions are being made," said Fletcher. Reid echoed the point. Many chief diversity officers, she said, talked about the money they spend on diversity programs. "But they don't feel they have a big voice in technology decisions. We're trying to weave in and embed inclusive talent-management practices," she said. "How do we recognize and interrupt when a biased decision will be made?"

An important strategy to eliminate bias, the panelists agreed, is by harnessing technology like machine learning. "Diversity officers need to think—technology first," said Burlacu. "How can technology help them in how they work as an organization?"

Fletcher said she's hearing from businesses who want to use analytics to figure out, "Where do we stand?" The primary challenge, she said, is determining the goals of the company.

SEE: 10 tools to help your company improve diversity

It's no secret that bias is baked into machine learning algorithms. Why? Because these algorithms are dependent on large sets of data—which, in itself, represents bias.

"If I Google CEO, the first 32 images are of men," said Baur. "Why is that? Google is not biased. Our data is biased." The same logic applies to organizations, she said. "If I'm in an organization that has been biased, machine learning will reliably come up with predictions for males to be promoted. You need to work deliberately on the data and in some cases, intervene with machine learning and de-bias the algorithm."

Also see...

About Hope Reese

Hope Reese is a Staff Writer for TechRepublic. She covers the intersection of technology and society, examining the people and ideas that transform how we live today.

Editor's Picks

Free Newsletters, In your Inbox