Which of the following are listed as causes of algorithmic bias?

• A. Historic human biases
• B. Incomplete and unrepresentative training data
• C. Lack of representation among people designing and developing algorithms
• D. All of the above

Answer: (D)

Algorithmic bias comes from multiple intertwined sources. Historic human biases are embedded in the data and decisions that feed systems, shaping outcomes in subtle and sometimes harmful ways. Training data that is incomplete or not representative of the real-world population can cause a model to perform well for some groups while misfiring for others. When the people designing algorithms aren’t diverse, biases can creep in through choices about what to measure, how to frame problems, and which safeguards to implement. Because each of these factors can contribute to biased results, the best answer is that all of these are causes. In practice, addressing bias involves using representative data, auditing performance across groups, and ensuring diverse teams participate in design and evaluation.