Based in Vancouver, Canada, this is a blog by Michael Argast. My areas of interest include travel, economics, technology, the environment, personal development and other eclectic topics.

Programming bias...

Machine learning, the creation of black box algorithms using complex, data-intensive structures, is growing at a rapid pace.

Millions of jobs will be destroyed, industries will rise and fall. There is the possibility for great good to come out of this - algorithms that detect cancer, self-driving cars that greatly reduce auto deaths, make better movie recommendations.

"Algorithms don't make things fair, they repeat our past practices."

Unfortunately, we have biases embedded throughout society. If we look at past indicators of success, we often ignore the privilege or barriers that some had or had to overcome. If we use algorithms to pick job applicants, grant credit, design our cities, predict criminality - we will be embedding unconscious bias into these systems and reinforcing the unfairness of the existing system.

These systems will be programmed with people who often aren't even aware of their own biases, or who possibly believe that a bias is 'fair' because of a personal perspective. 

We still have significant societal challenges with bias. Most of us have biases we don't even recognize. We need to be very careful, and thoughtful, about how we pass these biases onto our systems, and then trusting those systems.

Further viewing:

Cognitive Biases

Review: The Shady Tree pub

Management Expense Ratios (financial advice)