The second statement makes no sense as support for the first.

The second statement is an explanation for why the datasets are flawed!

Bias is due to flawed datasets, which unfortunately are extremely difficult to avoid due to the myriad social inequalities whose impacts are embedded within all real-world data of non-negligible scope.