AI0-001 AI Governance and Ethics — 30 Questions

Question 1

A healthcare AI system uses patient data to predict disease risk. To comply with HIPAA and reduce the risk of re-identification, which technique should be applied to the training data before model development?

Accepted Answer

Differential privacy with a carefully chosen epsilon. Differential privacy (Option C) is the correct technique because it adds calibrated noise to the training data or model outputs, providing a mathematical guarantee against re-identification even if an attacker has auxiliary information. This directly addresses HIPAA's requirement to protect patient privacy while preserving statistical utility for disease risk prediction.

Answer

Pseudonymisation by replacing patient names with random IDs

Answer

Data augmentation to create synthetic samples

Answer

Data minimisation by removing all features except age and gender

Question 2

An AI team is developing a model that approves loan applications. The dataset contains historical loan decisions where a protected group was disproportionately denied loans. The team wants to ensure the model does not perpetuate this bias. Which fairness metric should be used during validation to directly measure whether the model's positive prediction rate is equal across groups?

Accepted Answer

Demographic parity. Demographic parity requires the probability of a positive prediction (loan approval) to be equal across groups. This directly addresses the concern of perpetuating historical denial rates. Equalised odds measures error rates, not positive prediction rates.

Answer

Calibration

Answer

Individual fairness

Answer

Equalised odds

Question 3

A company is deploying an AI system that screens job applications. According to the EU AI Act, this system is likely classified as high-risk because it affects employment opportunities. Which requirement must the company implement for high-risk AI systems?

Accepted Answer

A human-in-the-loop mechanism that enables override of the AI's decisions. The EU AI Act requires human oversight for high-risk AI systems to allow operators to override or stop the system's decisions when necessary. The other options are not mandated by the Act for high-risk systems.

Answer

Full transparency by publishing the model's source code and training data

Answer

Annual third-party audits of the model's energy consumption

Answer

Obtaining explicit consent from each applicant to process their data

Question 4

A data scientist is using SHAP to explain a complex ensemble model's predictions. A business stakeholder asks why a particular prediction was made. The data scientist wants to show the most influential features for that single prediction. Which SHAP visualisation is most appropriate?

Accepted Answer

A SHAP force plot for the individual prediction. A SHAP force plot is specifically designed to visualize the contribution of each feature to a single prediction, showing how features push the prediction from the base value (average model output) to the final prediction. This makes it the ideal choice for explaining an individual prediction to a business stakeholder, as it provides a clear, localized explanation of feature impacts.

Answer

A SHAP summary plot showing mean absolute SHAP values across all features

Answer

A SHAP dependence plot for the top feature

Answer

A SHAP bar chart of absolute feature importance

Question 5

A financial institution needs to deploy a credit scoring model that is interpretable to regulators. The model must provide clear reasons for each decision. Which model type should the institution choose?

Accepted Answer

A glass-box model such as logistic regression or a decision tree. Glass-box models like logistic regression or decision trees are inherently interpretable and can provide clear, auditable reasons for each prediction. Black-box models require post-hoc explainability methods, which may not be sufficient for regulatory scrutiny.

Answer

A gradient-boosted tree ensemble with SHAP explanations

Answer

A black-box model with a model card describing its behavior

Answer

A deep neural network with LIME explanations

Question 6

A company is training a large language model from scratch and wants to minimise its environmental impact. Which practice aligns with green AI principles?

Accepted Answer

Use model pruning and train on a smaller, representative dataset. Green AI advocates for resource-efficient AI, including using smaller models, pruning, and efficient architectures to reduce carbon footprint. Training larger models with more data increases environmental impact, not reduces it.

Answer

Use more GPUs to parallelise training and reduce wall-clock time

Answer

Deploy the model on a cloud provider with renewable energy certificates

Answer

Train the model on a larger dataset to improve accuracy

Question 7

An organisation is developing an AI policy. According to the NIST AI RMF, which function involves establishing policies and procedures to ensure the organisation governs AI responsibly?

Accepted Answer

Govern. The NIST AI RMF's Govern function focuses on establishing governance structures, policies, and accountability mechanisms. Map, Measure, and Manage are other functions in the framework.

Answer

Manage

Answer

Measure

Answer

Map

Question 8

A company uses AI to generate marketing images. They want to ensure that the images are clearly identified as AI-generated to comply with transparency obligations. Which approach is most effective?

Accepted Answer

Embed a visible watermark stating 'AI-generated' in each image. Watermarking AI-generated content with visible or invisible markers is a direct way to disclose AI origin. Using metadata alone can be stripped; disclaimers in terms of service are not visible to users; relying on human detection is unreliable.

Answer

Add a disclaimer in the platform's terms of service

Answer

Include metadata in the image file indicating it is AI-generated

Answer

Rely on deepfake detection algorithms to flag the images

Question 9

A hospital wants to train a diagnostic model using data from multiple hospitals without sharing raw patient data. Which technique allows model training across decentralised data while preserving privacy?

Accepted Answer

Federated learning. Federated learning trains a shared model by aggregating updates from local data without moving the data itself, preserving privacy. Differential privacy adds noise but doesn't decentralise data. Data centralisation violates privacy. Anonymisation alone doesn't allow collaborative training.

Answer

Differential privacy applied to the combined dataset

Answer

Centralising all data in one location and anonymising it

Answer

Using pseudonymisation and then pooling the data

Question 10

Under the GDPR, individuals have the right to not be subject to a decision based solely on automated processing if it produces legal effects. Which of the following is a typical safeguard that organisations must provide to comply with this right?

Accepted Answer

The right to obtain human intervention on the part of the controller. GDPR requires human intervention as a safeguard for automated decisions with legal effects. The right to explanation and human oversight are key. Deleting data is not a safeguard; consent is not sufficient; detailed explanations are part of transparency.

Answer

The right to have all personal data deleted immediately

Answer

The right to receive a detailed mathematical explanation of the model

Answer

The right to demand a more favourable automated decision

Question 11

A company is evaluating a vendor's AI system for hiring. The vendor claims the system is fair because it achieves demographic parity. However, the company discovers that the system has significantly different false positive rates across groups. Which fairness issue does this indicate?

Accepted Answer

The system violates equalised odds. Equalised odds requires that a model's false positive rates and true positive rates are equal across all demographic groups. Since the vendor's system has significantly different false positive rates across groups, it violates the equalised odds fairness criterion, even if demographic parity (equal selection rates) is satisfied. This is a core fairness metric in AI governance, as it ensures that errors are distributed equitably.

Answer

The system violates individual fairness

Answer

The system suffers from selection bias

Answer

The system is not calibrated

Question 12

An AI system trained on historical medical records shows that certain racial groups have higher predicted risk for a disease. The data reflects real-world differences in diagnosis rates due to unequal access to healthcare. Which type of bias is this?

Accepted Answer

Historical bias. Historical bias occurs when the training data reflects past societal inequities. Selection bias would occur if the sample is not representative; confirmation bias stems from the model's own predictions; algorithmic bias arises from the model design.

Answer

Algorithmic bias

Answer

Selection bias

Answer

Confirmation bias

Question 13

A company is implementing an AI ethics board. Which TWO responsibilities should the board typically have?

Accepted Answer

Approving or rejecting high-risk AI initiatives. An AI ethics board should oversee AI projects for ethical compliance and approve high-risk AI initiatives. Implementing technical models and auditing data privacy are operational tasks not typically under the board's direct purview.

Answer

Writing code for fairness algorithms

Answer

Developing marketing strategies for AI products

Answer

Conducting daily data privacy audits

Question 14

A company is deploying an AI system that falls under the EU AI Act's high-risk category. Which THREE requirements must the company fulfill?

Accepted Answer

Ensure human oversight to prevent or minimise risks. The EU AI Act for high-risk systems requires risk management, human oversight, and transparency documentation. Open-sourcing the model is not required; obtaining consent is not a specific requirement for high-risk systems.

Answer

Obtain explicit consent from all affected individuals

Answer

Open-source the model's code to the public

Question 15

A data scientist is using LIME to explain a black-box model. Which TWO characteristics of LIME are true?

Accepted Answer

It creates an interpretable surrogate model locally around a prediction. LIME (Local Interpretable Model-agnostic Explanations) creates local surrogate models to explain individual predictions. It does not provide global feature importance or an inherent measure of model confidence.

Answer

It provides a measure of model confidence in its prediction

Answer

It requires access to the model's internal parameters

Answer

It provides a global ranking of feature importance across the entire dataset

Question 16

A healthcare AI system diagnosing diabetic retinopathy from retinal images shows high accuracy overall but significantly lower recall for patients with darker skin tones. Which fairness metric would BEST capture this disparity by comparing true positive rates across groups?

Accepted Answer

Equalised odds. Equalised odds requires that the true positive rate and false positive rate be equal across groups. Demographic parity only checks outcome rates, not error types. Individual fairness compares similar individuals. Calibration checks confidence alignment.

Answer

Calibration

Answer

Demographic parity

Answer

Individual fairness

Question 17

A data scientist needs to explain a single prediction from a complex ensemble model to a business stakeholder. Which technique generates local, interpretable explanations by perturbing input features and fitting a simple surrogate model?

Accepted Answer

LIME. LIME (Local Interpretable Model-agnostic Explanations) explains individual predictions by perturbing inputs and learning a linear surrogate. SHAP provides Shapley values, which are also local but game-theoretic. Attention visualisation is for transformer models. Model cards describe global model behaviour.

Answer

Model card

Answer

SHAP

Answer

Attention visualisation

Question 18

A company deploys an AI resume screening tool. It learns from historical hiring data where most successful hires were male, leading the model to favour male candidates. Which type of bias is this primarily?

Accepted Answer

Historical bias. The model learned from historical hiring data that already contained a gender imbalance, where most successful hires were male. This is a classic case of historical bias, where the training data reflects past societal or organizational biases, and the AI system perpetuates those biases in its predictions. The bias originates in the data, not in the model's algorithm or the sampling method.

Answer

Confirmation bias

Answer

Algorithmic bias

Answer

Selection bias

Question 19

Under the EU AI Act, an AI system that uses subliminal techniques to materially distort a person's behaviour, causing psychological or physical harm, would be classified under which risk tier?

Accepted Answer

Unacceptable risk. The EU AI Act categorises such systems as 'unacceptable risk' and prohibits them outright. High risk includes critical infrastructure, education, employment, etc. Limited risk involves transparency obligations. Minimal risk covers all other systems.

Answer

Limited risk

Answer

High risk

Answer

Minimal risk

Question 20

A hospital wants to train a diagnostic model using patient data from multiple hospitals without sharing raw patient records. Which technique enables collaborative model training while keeping data decentralised?

Accepted Answer

Federated learning. Federated learning trains models across decentralised data without exchanging raw data. Differential privacy adds noise. Anonymisation removes identifiers. Pseudonymisation replaces identifiers with pseudonyms.

Answer

Pseudonymisation

Answer

Differential privacy

Answer

Anonymisation