Why aren't algorithms impartial?

Question $19$ options:

The datasets that they rely on often only contain specific populations, white and male or Muslim and male, for instance

The data used to train them, as in facial recognition software, could have its own biases based on its historical use.

Those who program the algorithms in the first place may only come from one social group end embed their own unconscious biases into the code.

all of these

none of these