Techstars DC Startup Week

“A young child defines the world purely on the small amount they can see…This is the root of dataset bias: intelligence based on information that’s too small or homogenous.” Advances in AI technologies like machine learning and deep neural networks have potential to save time and boost productivity. But what if we train these technologies using datasets that exclude large portions of the population?
For example, some facial recognition software doesn’t acknowledge dark skin. Why? People of color were excluded from the datasets that were used to train the software. If AI isn’t designed with inclusion upfront, its rewards won’t equally benefit us all. This talk shares the risk of unconscious bias in algorithmic datasets, and how developers can overcome it.