> It's well known that neural networks can easily inherit biases from their training data. It's also well known that datasets generated by western universities

> It's well known that neural networks can easily inherit biases from their training data. It's also well known that datasets generated by western universities are widely used in training and evaluating neural networks. If my training set is full of pictures of Stanford CS undergraduates, I could end up with a computational photography system that makes everyone look like Stanford CS undergraduates, or a historical photo colourisation system that makes everyone look like Stanford CS undergraduates, or a self driving car pedestrian tracking system that expects 90% of pedestrians to look like Stanford CS undergraduates. Why then aren't foreign universities/companies simply... building their own datasets?