Ghost in the machine: The unintended consequences of bias in machine learning

Description

From recommendations engines to deep learning algorithms to help detect cancer, machine learning is quickly becoming a normal part of our lives. Which means more possibilities of encountering algorithms gone awry—such as the infamous Tay the racist Twitter Bot. Every step of building an algorithm introduces risks for injecting unintended bias; yet we rarely focus on ways to mitigate or eliminate these. Instead, we accept these as the cost of operating in this space.

Nivia Henry walks you through the journey of how an algorithm is built; risk points for errors including algorithmic bias; and steps you can take to catch, reduce, or eliminate them before they cause harm to your users.