Fascinating new research finds that a certain type of algorithm can make better hiring decisions than humans.
The algorithm is designed to favor “rare” candidates — say, a woman in an applicant pool that’s mostly men. The algorithm also learns quickly from its decisions. So if that woman ends up getting hired and performing well, the algorithm will start selecting more candidates like her.
Compared to the people human hiring managers select, the algorithm identifies higher-quality talent and increases diversity. It’s even more effective than standard hiring algorithms, which require you to tell them who’s been successful in the past so it can select more people like that.
What interests me here isn’t the fancy technology. It’s the broader idea that employers don’t always know what a promising job candidate looks like. And that taking chances on candidates who seem different from the people you typically hire can pay off.
I spoke to Danielle Li, an associate professor at MIT’s Sloan School of Management and a coauthor on the paper. She told me that standard hiring algorithms (the ones that require you to tell them who’s been successful in the past) assume that “any applicant we might encounter in the future, we’ve already encountered in the past and we know how that person’s going to do.”
We don’t.
The most successful employers in the future economy will stay curious about what makes a top performer. They’ll be willing to adjust their hiring process based on what they learn — just like they expect their employees to adapt and develop over time.
