AI no shortcut to diverse hiring
By AdvocateDaily.com Staff
The CBC recently reported on a tool developed by a Toronto start-up Knockri, which aims to remove unconscious bias from the early screening process used by employers when filling vacancies.
According to the news outlet, the software uses artificial intelligence to draw up a shortlist of the most suitable applicants by analyzing facial and speech responses in videotaped interview answers. The best candidates are ranked according to scores for factors such as confidence and collaboration and presented to employers without names or faces attached.
The company’s CEO Jahanzaib Ansari told the CBC he was inspired by his own experience job hunting when he found he was more likely to get called for an interview by replacing his first name with alternatives such as “Jason” or “Jay.”
But Rodney, the founder of Rodney Employment Law, says employers can not use artificial intelligence as a shortcut to improve diversity. To get the most out of the tool, he says organizations need to lay some groundwork first.
“It needs to be aligned with the organization’s culture. This kind of tool will work better in a company that is in a state of readiness, and has already taken meaningful steps to embrace diversity,” he says.
“If a firm is using artificial intelligence to give the appearance of being less biased, then it might give them some extra protection on the surface, but at the end of the day, it won’t make much difference in practice, because they’re not taking the issue seriously.
“Without that foundation, it doesn’t matter what you use because the bias remains,” Rodney adds. "Another potential challenge is the bias that can arise from those who design the software and program the data, and this would need to be carefully scrutinized in the up-front stages."
Still, he says the software shows great promise for employers who are willing to make a genuine commitment to diversity in hiring.
“We always advise clients when they’re recruiting to cast the net as wide as they can, and then filter people out,” Rodney says. “The most time-consuming part of any process comes at the top of that funnel, so there’s no question AI could improve efficiency by helping to filter out potential candidates.”
However, he worries that something could be lost in a recruitment process that relies so heavily on artificial intelligence and algorithms.
“There could be limitations in terms of measuring intangibles. For example, a computer looking at a long gap in a resume might mark it down as a red flag, whereas I might look at a person who has done a great deal of travelling as a great learning experience,” Rodney says.
In addition, he says it is likely to take a great deal of up-front investment to program software for the particular requirements of individual vacancies. For that reason, Rodney says employers with multiple identical roles would get the most out of it.
“You have to look at the long-term return on the short-term investment,” he says. “If you’re hiring hundreds of people to work in a call centre, where you can nail down the skills required to do the job, that’s going to a much better proposition than if you’re looking to fill a very niche position or one that is unique to the organization.”