Which of the following statements is WRONG according to the reading material $($part of The Ethical Algorithm$)$ by Michael Kearns and Aaron Roth?

Question $4$ options:

Restricting the inputs to algorithmic decision$-$making process can effectively eliminate the bias embedded in data.

Algorithms won't correct the bias unless humans explicitly ask for it$.$

Training data used in machine learning applications often contain hidden biases.

An alternative approach to tackle algorithmic bias is to define fairness of the model outputs.