What does the term "bias" mean in the context of machine learning?

Correct answer: Systematic errors or prejudice in model predictions that disadvantage certain groups

Option B is correct because in machine learning, bias refers to systematic errors or prejudice embedded in a model's predictions, often arising from unrepresentative training data, flawed labeling, or historical inequities, which can cause the model to consistently disadvantage or misrepresent certain demographic groups. Option A is incorrect because the number of parameters describes model capacity or size, not bias. Option C is incorrect because training time is a computational performance metric entirely unrelated to the statistical or ethical concept of bias. Option D is incorrect because memory required for model weights is a hardware resource consideration, not a measure of prediction fairness or systematic error.