Sign In

Communications of the ACM

ACM TechNews

Are Hiring Algorithms Fair? Study Finds They're Too Opaque to Tell


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
Hiring algorithms choose some candidates over others, but we're not sure why.

A study by Cornell University researchers raises doubts about algorithms' ability to make fair decisions when screening potential hires.

Credit: Harvard Business Review

A study by Cornell University researchers raises doubts about algorithms' ability to make fair decisions when screening potential hires.

The researchers found the makers of such algorithms prefer to keep their design and workings hidden.

Few vendors provided tangible data on how they validate algorithmic pre-employment screenings, or specified their bias mitigation strategies.

The researchers said vendors' claims that their algorithms are "fair" also can be vague, as they do not have to disclose the company's definition of fairness.

Cornell's Manish Raghavan said, "The real question is not whether algorithms can be made perfect; instead, the relevant comparison is whether they can improve over alternative methods, or in this case, the human status quo."

From Cornell Chronicle (NY)
View Full Article

 

Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA


 

No entries found