How to Keep Human Bias Out of AI

Author: Kriti Sharma

Publisher: TED

Publication Year: N/A

Summary: The following video features Kriti Sharma who talks about how human-programmed artificial intelligence (AI) can have the biases of humans encoded into them and how this can negatively affect real people. Even in everyday technology, having female voice assistants, such as Siri or Alexa, perpetuates sexism. AI is another piece of technology that we encounter regularly, in curated online ads, etc. However, the same AI technology profiles women to see pregnancy test ads and profiles Black men to be more likely to commit a crime. This is problematic, but AI is within our control and therefore we can fix these biases by accounting for 3 things: 1). Being aware of our own biases that we may encode into AI; 2). Having diverse teams build AI; and 3). Giving the AI diverse experiences to continue to learn from. Taking these steps can help decrease problematic bias within human-programmed AI.