Discrimination can’t be easily removed when programs are created based on past practices
“We’re trying to fix something that is broken,” said Lindsey Zuloaga, chief data scientist at HireVue Inc., one of many companies offering AI solutions to HR problems. “From the beginning, we felt the resumé was not a good representation of a person.”
For example, time gaps in a resumé can often be seen as a negative by a recruiter, even if those gaps are easily explained.
The process of crafting a cover letter and curriculum vitae can also present accessibility barriers to some individuals. And with more job opportunities requiring new and dynamic skill sets, past performance, particularly in positions more than 10 years old, isn’t a good benchmark for evaluating a candidate.
“This old way of doing this is not going to work,” Zuloaga said. “We have to know how skills transfer.”
We’re trying to fix something that is broken
Lindsey Zuloaga
Video interviews, interactive assessments and virtual job tryouts are better indicators of future success in a role, HireVue said, than experience listed on a resumé.
The increased implementation of new technologies coincides with companies placing more value on future potential than past experiences.
“Language models are really good at understanding relationships between skills,” Zuloaga said.
The technology should also make the process simpler and more practical for applicants, proponents say. For example, instead of applying for several job openings at a company, it can keep your information on file and connect with you when openings arise.
Companies, of course, often tell unsuccessful applicants that their resumé will be kept on file, but there’s likely not a system in place to cull that data or wield it effectively. AI, however, can turn that promised database into a reality and better serve job seekers than anything that currently exists.
Chatbots can also be used to help connect applicants with openings, answering questions and assisting people in the process.
“We’re hoping to get away from this old model of applying for jobs,” Zuloaga said. “My vision of the future is more candidate-centric.”
The way new programs and approaches look at applicants, she said, “offers more data than a resumé or LinkedIn profile.” For workers seeking job stability and long-term growth, companies are hoping an improved application process can also reduce turnover.
In addition, 75 per cent of workers oppose AI making final decisions, although half of those surveyed believe in the potential for AI to reduce bias in hiring.
Discrimination is always going to creep into the ways you build the algorithm
Valerio de Stefano
“Discrimination is always going to creep into the ways you build the algorithm,” Valerio de Stefano, a professor and Canada Research Chair in Innovation Law and Society at York University, said. “Removing bias is a secondary objective,” as the main goal of these programs is to help companies cut costs. “If you have to sift through thousands of CVs, you can save a lot of money if you outsource to a program.”
Zuloaga said humans can’t rid themselves of bias, but it can be deleted from a program.
“You can remove information from AI,” she said, citing the importance of transparency. “Having structure and standardization is important. We remove demographic data and bring in mathematical techniques to predict accuracy and maximize fairness.”
But De Stefano believes discrimination can’t be so easily removed, in part because programs are likely created based on past discriminatory practices.
“In order to build any of these programs, you have to have a database and a benchmark that informs how they are going to work,” he said. “If that database or dataset includes discrimination — and, historically, we have discrimination in many jobs — it’s not so easy to root out.”
De Stefano doesn’t think any companies set out to discriminate, but noted that even a genuine struggle to remove biases from such programs can’t be done.
“The idea that we can use technology as a magic wand to eliminate discrimination is just delusional,” he said. “Tech is a reflection of what we do and value.”