Avoiding biased algorithms: Lessons from the hiring space

Description

There have been many headlines of algorithms gone wrong: racist bots, failing financial markets, and unfair practices in law enforcement, loan approval, and hiring. Not only are these outcomes harmful; they’re a PR nightmare for the people who built the algorithms behind them. So how can these situations be avoided? Lindsey Zuloaga shares experiences and lessons learned in the hiring space to help others prevent unfair modeling and explains how to establish best practices. Lindsey demonstrates how, contrary to these examples, AI can actually help us overcome bias and improve diversity and explains how her team uses machine learning algorithms to overcome implicit and explicit bias in hiring.

To give one example, algorithms that predict job performance from video interviews have input features that make them essentially blind to many of the characteristics humans unfairly use to evaluate others (age, race, gender, etc.). Still, as we have seen in many of the examples of algorithms gone wrong, features that are correlated to a protected class may persist. In these cases, there are systematic ways to isolate and mitigate bias. These practices should be standard in any situation where algorithms are being used to score or evaluate people.