We agree. Dataset fidelity and bias are major concerns for publicly available datasets. For this reason we are working to develop programmatically created datasets along with anti-bias testing and policies.