DSE Ethical Considerations in AI 3 — Questions and Answers
Question 1: The concept of 'informed consent' in AI data collection requires that participants:
- Sign any legal document provided by the organization
- Understand how their data will be used before agreeing (Correct answer)
- Provide data anonymously by default
- Only consent once per organization
Correct answer: Understand how their data will be used before agreeing
Informed consent means individuals must clearly understand the purpose and use of their data before agreeing to share it.
Question 2: Which of the following best describes 'model cards' as an ethical tool?
- Credit scoring models for financial institutions
- Standardized documents describing a model's intended use, performance, and limitations (Correct answer)
- Cards that store encrypted model weights
- API tokens for accessing ML models
Correct answer: Standardized documents describing a model's intended use, performance, and limitations
Model cards are structured documentation artifacts that communicate a model's purpose, performance metrics, and known limitations across subgroups.
Question 3: When a predictive policing algorithm disproportionately targets a specific neighborhood due to historically biased arrest data, this illustrates:
- Feedback loop amplification of bias (Correct answer)
- Correct use of base rate statistics
- Overfitting to training data
- Underfitting due to sparse data
Correct answer: Feedback loop amplification of bias
Biased predictions lead to increased policing in certain areas, generating more data that reinforces the bias in a self-perpetuating feedback loop.
Question 4: Under GDPR, the 'right to erasure' (right to be forgotten) means:
- Individuals can demand deletion of their personal data under certain conditions (Correct answer)
- All EU data must be deleted after 5 years automatically
- Only government data must be erased on request
- Companies must erase data after each model training run
Correct answer: Individuals can demand deletion of their personal data under certain conditions
GDPR's right to erasure allows individuals to request deletion of their personal data when it is no longer necessary or consent is withdrawn.
Question 5: Counterfactual explanations in AI ethics are used to:
- Detect adversarial attacks on models
- Show what minimal changes to input would alter a model's decision (Correct answer)
- Generate synthetic data for fairness testing
- Audit training pipelines for data leakage
Correct answer: Show what minimal changes to input would alter a model's decision
Counterfactual explanations tell users 'if X had been different, the outcome would have changed,' making decisions more actionable and understandable.
Question 6: A data scientist omits race as a feature but includes zip code, which correlates strongly with race. This practice is known as:
- Feature ablation
- Proxy discrimination (Correct answer)
- Dimensionality reduction
- Label smoothing
Correct answer: Proxy discrimination
Proxy discrimination occurs when a seemingly neutral variable acts as a stand-in for a protected attribute, reintroducing the discrimination being avoided.
Question 7: Which ethical framework emphasizes maximizing overall benefit to society when evaluating AI deployment decisions?
- Deontological ethics
- Virtue ethics
- Utilitarian ethics (Correct answer)
- Contractarian ethics
Correct answer: Utilitarian ethics
Utilitarianism evaluates actions by their consequences, favoring decisions that produce the greatest good for the greatest number.
DSE Ethical Considerations in AI 3
The concept of 'informed consent' in AI data collection requires that participants: