DSE - Data Science Ethical Considerations in AI Questions and Answers — Questions and Answers
Question 1: A bank uses an AI model for loan approvals. It's discovered that the model denies loans to a disproportionately high number of applicants from a specific zip code, which is strongly correlated with a protected demographic group. Even though applicants' financial profiles are similar to approved applicants from other areas, their applications are rejected. What is the most likely ethical issue at play?
- Lack of model accuracy
- Insufficient training data volume
- Algorithmic bias (Correct answer)
- Poor feature engineering
Correct answer: Algorithmic bias
This scenario describes algorithmic bias, where an AI system produces systematically prejudiced outcomes due to erroneous assumptions in the machine learning process. In this case, the zip code is acting as a proxy for a protected attribute (e.g., race), causing discriminatory outcomes even if the protected attribute itself is not used.
Question 2: Which of the following best describes the primary goal of Explainable AI (XAI)?
- To achieve the highest possible prediction accuracy regardless of model complexity.
- To allow stakeholders to understand, trust, and manage the results of an AI model. (Correct answer)
- To reduce the computational cost and time required for training complex models.
- To fully automate the process of feature selection and data preprocessing.
Correct answer: To allow stakeholders to understand, trust, and manage the results of an AI model.
The core purpose of Explainable AI (XAI) is to make the decision-making process of AI models, especially complex 'black-box' models, transparent and understandable to humans. This fosters trust, ensures accountability, and allows for effective auditing and debugging.
Question 3: An organization wants to train a machine learning model on sensitive user data while providing strong mathematical guarantees that the model's output will not reveal information about any single individual in the dataset. Which technique is specifically designed for this purpose?
- Differential Privacy (Correct answer)
- K-Anonymity
- Data Encryption
- Feature Hashing
Correct answer: Differential Privacy
Differential Privacy is a formal framework that provides strong, mathematically provable guarantees of privacy. It works by adding precisely calibrated statistical noise to data or query results, making it impossible to determine whether any single individual's data was included in the dataset, thus protecting individual privacy while allowing for useful aggregate analysis.
Question 4: A data science team is deploying a critical AI system for medical diagnosis. To ensure accountability for the model's predictions, which of the following is the MOST crucial practice to implement?
- Using the most complex deep learning algorithm available for the task.
- Anonymizing all patient data before the training process begins.
- Maximizing the model's raw accuracy above all other evaluation metrics.
- Establishing a clear governance framework with version control, decision-logging, and designated human oversight. (Correct answer)
Correct answer: Establishing a clear governance framework with version control, decision-logging, and designated human oversight.
Accountability in AI requires the ability to audit, understand, and assign responsibility for a system's outcomes. A governance framework with detailed logging, model versioning, and clear lines of human oversight is essential for this. While data anonymization is vital for privacy, it does not address the accountability of the model's diagnostic decisions.
Question 5: In the context of AI fairness, what does "demographic parity" (also known as statistical parity) require?
- The model must have the same true positive rate and false positive rate across all demographic groups.
- The proportion of individuals receiving a positive outcome must be the same across different demographic groups. (Correct answer)
- The model must achieve the highest possible accuracy for all groups, even if it leads to different selection rates.
- The model's predictions must be fully explainable to individuals from all demographic backgrounds.
Correct answer: The proportion of individuals receiving a positive outcome must be the same across different demographic groups.
Demographic parity is a group fairness metric that is satisfied if the likelihood of a positive outcome is the same across different protected groups (e.g., race, gender). It focuses solely on making the prediction rates equal, independent of the true outcomes.
Question 6: A company implements a "human-in-the-loop" (HITL) system for its AI-powered content moderation platform. What is the PRIMARY ethical advantage of this approach?
- It significantly reduces the computational cost of the AI model.
- It fully automates the process of retraining the model on new data.
- It provides a mechanism for human oversight to correct errors and handle nuanced cases the AI might misinterpret. (Correct answer)
- It guarantees the complete elimination of all forms of bias from the system.
Correct answer: It provides a mechanism for human oversight to correct errors and handle nuanced cases the AI might misinterpret.
The main benefit of a human-in-the-loop system is combining the speed and scale of AI with the nuance, context, and ethical reasoning of human judgment. This allows humans to intervene, correct errors, and handle ambiguous or high-stakes decisions that an automated system might get wrong, which is crucial for ethical applications like content moderation.
DSE - Data Science Ethical Considerations in AI Questions and Answers
A bank uses an AI model for loan approvals.
It's discovered that the model denies loans to a disproportionately high number of applicants from a specific zip code, which is strongly correlated with a protected demographic group.
Even though applicants' financial profiles are similar to approved applicants from other areas, their applications are rejected.
What is the most likely ethical issue at play?