Can resume-reading software help companies make better hires?
While some fear artificial intelligence may take jobs from humans, technology company SAP sees it as a way to potentially make better hires and increase diversity.
The German technology company, which employs more than 1,500 people in the Chicago area, is introducing a tool that will allow recruiters to use machine learning to sift through thousands of applications much faster.
SAP Resume Matching applies machine learning to the process of matching resumes and job descriptions, said DJ Paoni, SAP’s Chicago-based Midwest managing director. SAP will roll the product out to its own recruiters this year and will also sell it to clients.
Paoni said the tool extracts information such as skills and experience from resumes and scores them against particular open positions. That can allow a recruiter to more easily whittle down a pool of a thousand to several dozen that are worth further consideration, he said.
“It really allows recruiters to focus on the important part of that whole process, which is interacting with the candidates, as opposed to poring through resumes and trying to match job descriptions,” Paoni said.
He said SAP plans to use this tool to help make better hires. Over the past eight years or so, the company has shifted to taking on more young, entry-level employees for the first time. As a result, it’s paying more attention to hiring and retention trends, such as the impact of employee well-being on productivity, diversity and inclusion, the use of part-time or supplemental workers and continuous feedback rather than annual reviews.
To make SAP a better workplace based on those trends, it needs to be quicker and better at finding the best hires, he said.
Automating the resume sorting process could also help remove bias from the hiring process, Paoni said, creating a more diverse candidate pool.
SAP also plans to use machine learning to score job descriptions themselves — a process that could help identify unconscious bias in listings. For example, terms like “rockstar” or “ninja” may be more attractive to men than women. The company is developing a new machine learning tool that would detect this kind of language using sentiment analysis, then suggest alternatives.
“It’s similar to a grammar check that you might do on a document,” Paoni said. “The system will recommend alternative words for terms that might hint at a pattern of unconscious bias.”
He said that will help SAP as it pushes for more diversity and inclusion. But Paoni said the machine learning tools are intended to speed up and supplement the recruiting process, not replace it.
“If you rely too much on the technology, you lose that personal feel,” he said. “It’s a delicate balance.”
This article is reprinted by permission from