Computing Bias - AP Computer Science Principles
- Identify a method to detect bias in datasets.
- Statistical analysis. Mathematical tests reveal unequal treatment across groups.
- What is the role of accountability in computing bias?
- Holds developers responsible for outcomes. Developers must answer for discriminatory algorithm outcomes.
- What is automation bias?
- Overreliance on automated systems. Humans trust computer decisions without proper scrutiny.
- What is anchoring bias?
- Relying too heavily on initial information. Initial information disproportionately influences final decisions.
- What is algorithmic bias?
- Bias arising from algorithm design and implementation. Occurs when the algorithm's structure inherently favors certain groups.
- What is implicit bias?
- Unconscious attitudes affecting decisions. Hidden prejudices influence seemingly objective decisions.
- Which bias arises from non-random data collection?
- Sampling bias. Occurs when data collection methods favor certain groups.
- What role does diversity play in reducing bias?
- Promotes varied perspectives and reduces bias. Different backgrounds challenge assumptions and identify blind spots.
- What is confirmation bias in computing?
- Favoring information confirming preconceptions. Developers design systems based on their existing beliefs.
- Identify a common cause of measurement bias.
- Inaccurate data collection tools. Sensors or surveys that systematically misrepresent reality.
- Identify the bias in historical data algorithms.
- Perpetuating outdated norms. Past discrimination gets encoded into present-day decisions.
- What is the significance of fairness in AI ethics?
- Ensures equitable treatment across groups. All groups should receive equal quality of service.
- Identify an ethical concern with computing bias.
- Unfair treatment of individuals or groups. Algorithms perpetuate discrimination against protected classes.
- Identify a legal concern related to computing bias.
- Violation of anti-discrimination laws. Biased systems can violate equal protection laws.
- Which term describes bias from unrepresentative samples?
- Selection bias. Samples don't accurately represent the broader population.
- Identify a form of bias in predictive policing.
- Racial profiling. Using location as proxy for race creates discriminatory outcomes.
- What is a feedback loop in computing bias?
- Biased outcomes reinforce biased data. Algorithm decisions create new biased data for future training.
- What is the role of transparency in reducing bias?
- Allows scrutiny and correction of biases. Open algorithms enable identification and correction of bias.
- Which type of bias is caused by flawed assumptions?
- Confirmation bias. Programmers' assumptions get embedded into the algorithm.
- Identify a potential bias in loan approval algorithms.
- Discrimination based on zip code. Location serves as proxy for protected characteristics.
- Which type of bias is present if an algorithm favors data trends?
- Trend bias. Algorithm learns patterns that may not apply universally.
- What is representational bias?
- Underrepresentation of certain groups in data. Some demographics are missing or undersampled in training data.
- Which bias is due to overfitting a model?
- Model bias. Models trained too specifically fail to generalize fairly.
- What is the impact of bias in automated decision-making?
- Potential for unfair outcomes. Biased systems create systematic disadvantages for certain groups.
- State a method to reduce bias in algorithms.
- Use diverse and representative datasets. Inclusive datasets prevent algorithms from learning harmful patterns.
- What is the impact of data imbalance on bias?
- Skews algorithm results. Unequal group representation distorts learned patterns.
- What is data bias?
- Bias introduced by data selection or processing. Problems in how data is collected, labeled, or preprocessed.
- What is a proxy variable?
- Variable used as a substitute for another. Indirect measure that may introduce unintended correlations.
- Identify a strategy to mitigate bias in AI systems.
- Regular audits for bias detection. Systematic testing helps identify discriminatory patterns early.
- Which type of bias results from algorithm simplifications?
- Overgeneralization bias. Algorithms make broad assumptions that miss important nuances.
- Which option is a consequence of biased algorithms?
- Discrimination. Biased algorithms systematically disadvantage certain groups.
- Which bias occurs when outcomes reinforce pre-existing biases?
- Feedback bias. System outputs create conditions that reinforce the original bias.
- What is exclusion bias?
- Excluding relevant data points. Important subgroups are left out of the analysis.
- What is the definition of computing bias?
- Systematic errors in algorithms affecting outcomes unfairly. These errors create unequal treatment across different groups.
- Identify one source of bias in computational data.
- Biased training data. Training datasets reflecting existing prejudices lead to biased models.