Algorithms are supposed to be neutral, but they are packed with the same biases built into the data used to create them. Wharton statistics professor James Johndrow has developed a method to remove those biases.
Source: Knowledge @ Wharton

Removing Human Bias from Predictive Modeling