When we hear stories about companies like Lockheed Martin using a "14-foot-tall, gleaming black box that can only be opened by solving one of the world's most impossible aerospace equations" as part of their hiring process, the future feels like it's finally arrived and our heads begin spinning with the possibilities of applying similar technology to our own recruiting and interview programs.
A few words of caution, however: federal and state antidiscrimination laws still apply to hiring, whether done by human interviewers or artificial intelligence. Just ask Amazon, which was forced to scrap its AI recruiting algorithm after discovering anti-women bias. So, before you hand over your hiring decisions to the machines, read this.
"Everyone wanted this holy grail," one of Amazon's machine-learning specialists told Reuters. "They literally wanted it to be an engine where I'm going to give you 100 resumés, it will spit out the top five, and we'll hire those." In theory, such an algorithm could remove any race, gender, or age bias out of the hiring process, and save companies from such complaints, while at the same time delivering the best candidates for the job. But that's not exactly what happened.
Because the company's computer models were trained to seek the best applicants based on patterns in resumés submitted to the company over a prior 10-year period -- and because most of those resumés came from men -- Amazon's AI taught itself that male candidates were preferable. Reuters reports the algorithm penalized resumes containing the word "women's" (i.e., "women's chess club captain") and also downgraded graduates of all-women's colleges. Engineers allegedly tried to re-train the AI to disregard gendered language, but it didn't take, and Amazon ditched the 3-year project last year. The company also contends the tool "was never used by Amazon recruiters to evaluate candidates," according to Reuters.
Small Biz and Big Data
The story of Amazon's sexist AI is a warning to other businesses trying to streamline the hiring process, even if their intent is to avoid any human bias. Cathy O'Neil at Bloomberg notes that Amazon is far from the only company utilizing algorithms to analyze job candidates. "What makes Amazon unusual," O'Neil writes, "is that it actually did its due diligence, discovered the troubling bias and decided against using the algorithm."
The essential problem is that, like Amazon's, many of these data analysis tools are relying on past practices and performances to predict future probabilities. And in male dominated industries -- like tech, finance, and law -- the algorithm reinforces those gender- and race-based hiring decisions, weeding out women and minorities rather than eliminating bias. And while businesses relying on AI may want "plausible deniability" according to O'Neil by hiding behind their hi-tech hiring, they could find themselves in big legal trouble.
So, before you let AI take over your HR, talk to an experienced employment attorney.