Which statement best describes bias in AI?

Which statement best describes bias in AI?

• A. Bias is only a hardware issue
• B. Bias arises from flawed data or algorithms
• C. Bias is always intentional
• D. Bias cannot be mitigated

Answer: (B)

Bias in AI shows up as systematic errors in predictions or decisions that can disadvantage certain groups. It often comes from the data the model learns from—if the data are unrepresentative, mislabeled, or reflect historical prejudice—or from how the model is designed or trained, such as the objective function that guides optimization. This makes the idea that bias arises from flawed data or algorithms the most accurate summary because it covers both data problems and modeling choices. Hardware issues don’t create this kind of bias, and bias isn’t always intentional—the model can learn biased patterns even without anyone aiming for unfair outcomes. Bias can be reduced through careful data collection and labeling, fairness-aware modeling, auditing, and ongoing monitoring.

Bias in AI shows up as systematic errors in predictions or decisions that can disadvantage certain groups. It often comes from the data the model learns from—if the data are unrepresentative, mislabeled, or reflect historical prejudice—or from how the model is designed or trained, such as the objective function that guides optimization. This makes the idea that bias arises from flawed data or algorithms the most accurate summary because it covers both data problems and modeling choices. Hardware issues don’t create this kind of bias, and bias isn’t always intentional—the model can learn biased patterns even without anyone aiming for unfair outcomes. Bias can be reduced through careful data collection and labeling, fairness-aware modeling, auditing, and ongoing monitoring.