In the context of artificial intelligence, bias refers to a system's tendency to make decisions or predictions that are prejudiced due to assumptions made during its development. This can result from the data used for training the AI, the way the algorithm is structured, or the subjective choices of the developers.