Author: Mike Bugembe
Publisher: TED
Publication Year: N/A
Summary: The following talk discusses how algorithms inherently have bias and there is a common problem of blind faith in data. Blind faith in data is trusting whatever comes from the data output because of our assumptions that data cannot be racist or sexist. The speaker gives an example of a young boy who was arrested wrongfully because face recognition thought he looked like someone else. These issues happen on an extremely large scope that greatly impacts people’s lives. If your data is not representative, for example, if the data you are using to build your model has a disproportionate amount of males to females this can lead to bias in the algorithms. This is where data storytelling can become helpful for data scientists so that they are not just trusting the data and they instead asking questions about the data. Adding a story will impact the decisions that are made. Stories are powerful and will reduce the blind trust the data scientists have and nearly everyone impacted by artiticial intelligence.