Artificial Intelligence needs enormous amounts of data to be trained. Currently, it is estimated that 15-20% of data used by data scientists is garbage and 80% of their time is spent scrubbing and cleaning the data. This means high-quality data is often hard to obtain, expensive and hard to scale to new markets. But one of the emerging challenges when it comes to data are the biased datasets due to the usage of wrong/disproportional sources and gaps on the data collection. There’s also a human intervention when it comes to data collection and building algorithms so they will, consciously or unconscientiously, reflect preconceptions, including cultural and socioeconomical backgrounds. In this talk and demo, we will discuss how to reduce bias in AI and how to get bias-free datasets that at the end will turn out into high-quality training data.