What is algorithmic bias, and which of the following is an example in content recommendation?

• A. Systematic skew in algorithm outputs due to training data or design; example: political content aligns with existing beliefs.
• B. The algorithm's outputs are always objective.
• C. Bias only arises in image processing.
• D. Bias can be fully eliminated by policies.

Answer: (A)

Algorithmic bias is when a system’s outputs reflect systematic errors or prejudices that come from the data it trained on or the design decisions behind it. In content recommendation, this shows up as the system pushing certain types of content based on past interactions, which can create a narrowing loop that reinforces existing preferences or beliefs. For instance, if political content that aligns with a user’s current views is shown more often, the recommendations can increasingly favor that content, making the feed feel biased toward a particular viewpoint. This illustrates bias embedded in how the algorithm operates, not a neutral snapshot of reality.

The other statements miss the mark because an algorithm’s outputs are not inherently objective, bias can show up across many domains beyond image processing, and while policies can help reduce bias, they cannot guarantee it is fully eliminated.