AP CSP Day 14: Bias In Data And Algorithms

Key Concepts

Algorithmic bias occurs when a model or decision system produces systematically unfair outcomes, often because the training data underrepresents certain groups. Data bias can stem from who collected the data, what was measured, or how samples were selected. AP CSP exam questions about bias ask students to identify potential sources of unfairness in a described algorithm or dataset. Recognizing that bias can be unintentional and that fairness requires deliberate effort in data collection and algorithm design is a key exam concept.

📚 Study the Concept First (Optional) Click to expand ▼

Bias in Data and Algorithms

What Is Algorithmic Bias?

Algorithmic bias occurs when a computational system produces outcomes that are systematically unfair to certain groups. Bias often enters through training data that overrepresents some groups and underrepresents others.

Real-World Examples

A facial recognition system trained mostly on lighter-skinned faces performs poorly on darker-skinned faces. A resume screening algorithm trained on historical hiring data perpetuates past hiring discrimination. Neither programmer intended harm, but the data encoded existing inequities.

Common Trap: Assuming that because a decision is made by an algorithm rather than a person, it is automatically fair. Algorithms reflect the biases present in their training data and design choices.
Exam Tip: AP exam bias questions often describe a real system and ask you to identify the source of bias. Look for who is underrepresented in the data collection process.
Big Idea 5: Impact of Computing
Cycle 1 • Day 14 Practice • Medium Difficulty
Focus: Bias in Data & Algorithms

Practice Question

A hiring algorithm is trained on data from a company that historically hired mostly men for engineering roles. Which of the following outcomes is most likely?

Why This Answer?

Machine learning algorithms identify patterns in training data. If the historical data shows that men were hired more frequently, the algorithm learns to associate male-related features with successful candidates. This reproduces and potentially amplifies the existing bias.

Why Not the Others?

A) Algorithms trained on biased data inherit that bias — they are not inherently objective. C) The algorithm considers all correlations in the data, including indirect demographic signals like names or activities. B) Algorithms do not self-correct for bias without explicit human intervention and redesign.

Common Mistake
Watch Out!

Students assume algorithms are inherently neutral because they are computer programs. In reality, algorithms reflect the biases present in their training data.

AP Exam Tip

Biased training data leads to biased algorithms. On the AP exam, if a question mentions historical data with known disparities, the algorithm will likely perpetuate those disparities.

Keep Practicing!

Consistent daily practice is the key to AP CSP success.

AP CSP Resources Get 1-on-1 Help
Back to blog

Leave a comment

Please note, comments need to be approved before they are published.