Mainak Mazumdar, a data scientist, discusses the need for better data sets for AI. This talk draws on real-life examples of poor data leading algorithms to bad decisions. This video communicates the reason AI projects need access to better, more diverse data. He identifies three infrastructure resets that need to happen to make AI ethical.  

Resource: Video  

Resource link: How bad data prevents good AI

Source credibility: TED is a non-profit, non-partisan organisation foundation dedicated to transmitting ideas.