AAIR Multiple Select — 20 Questions

Question 1

Which THREE actions should be taken when integrating AI risk into the broader ERM (Enterprise Risk Management) framework?

Accepted Answer

Establish aligned risk scoring criteria across all business functions.. Option B is correct because integrating AI risk into ERM requires a consistent, enterprise-wide risk scoring methodology so AI risks can be compared, aggregated, and prioritized alongside operational, financial, and compliance risks in the same register. Option D is correct because a shared risk taxonomy must be extended to capture AI-specific concepts such as model bias, data drift, hallucination, and model opacity; without common definitions, AI risks cannot be consistently identified, classified, or reported across functions. Option E is correct because AI risk owners must sit within the same governance and escalation channels (e.g., the same risk committee and reporting lines) as other risk owners, ensuring AI risk is not siloed and receives the same oversight, accountability, and board visibility as traditional risks. Option A does not belong because outsourcing all AI risk management removes internal ownership and accountability, which contradicts ERM principles of retained responsibility and integrated oversight. Option C does not belong because prohibiting every AI initiative with any inherent risk is an avoid-or-eliminate approach rather than risk-based management, and it would block innovation instead of treating, transferring, or accepting AI risk within defined tolerances.

Answer

Outsource all AI risk management to a third-party consultant.

Answer

Prohibit all AI initiatives that carry any inherent risk.

Question 2

Which TWO factors should be documented in an AI Model Card?

Accepted Answer

Intended use cases. Option D (Intended use cases) is correct because an AI Model Card is designed to document the purpose and appropriate applications of a model, so consumers know what tasks it was built for and where it should be deployed. Option E (Model limitations and known risks) is correct because Model Cards must disclose known biases, failure modes, out-of-scope scenarios, and other risks so users can make informed decisions about deployment. Options A, B, and C are incorrect because favorite snacks, stock prices, and new-hire handbooks are unrelated to model transparency, performance, or governance, and are not part of standard Model Card documentation such as that proposed by Mitchell et al. or required under emerging AI regulations.

Answer

The developer's favorite snacks

Answer

The current stock price of the company

Answer

The employee handbook for new hires

Question 3

Which THREE areas should be covered in an AI risk management program?

Accepted Answer

Model performance and accuracy. Option B (Model performance and accuracy) is correct because an AI risk management program must address model-specific risks such as drift, bias, overfitting, and degraded accuracy that directly affect decision quality and safety. Option C (Third-party/Vendor risk management) is correct because organizations often rely on external model providers, APIs, and training data suppliers, so vendor due diligence, contractual controls, and supply-chain monitoring are essential parts of AI risk governance. Option E (Legal and regulatory compliance) is correct because AI systems must comply with applicable laws and frameworks such as the EU AI Act, GDPR, and sector-specific regulations, making compliance a core risk domain. Option A (The office floor plan) is incorrect because physical office layout is a facilities concern, not an AI risk management area. Option D (The color of the company logo) is incorrect because branding aesthetics have no bearing on AI risk management.

Answer

The office floor plan

Answer

The color of the company logo

Question 4

Which TWO types of documentation are critical for AI Transparency requirements?

Accepted Answer

Data Sheets for Datasets explaining the provenance and makeup of training data.. Data Sheets for Datasets (A) are critical because AI transparency requires documenting the provenance, composition, collection methods, and preprocessing of training data so stakeholders can assess bias, consent, and suitability. Model Cards (B) are equally essential since they disclose a model's intended use, limitations, performance metrics across relevant subgroups, and ethical considerations, directly supporting transparency obligations. Together these artifacts give the technical and contextual information regulators and users need to understand how an AI system was built and how it should be used. The remaining options do not belong: a weekly meeting calendar (C) and an employee handbook (D) are general administrative documents unrelated to AI system transparency, and an office access list (E) concerns physical security, not AI documentation.

Answer

A weekly calendar of team meetings.

Answer

The employee handbook.

Answer

A list of all employees who have access to the office.

Question 5

The AI Risk Governance Committee is defining the risk appetite for a new generative AI chatbot. Which THREE factors must be considered to align with organizational risk tolerance?

Accepted Answer

Data privacy and residency requirements for training sets. Option B is correct because data privacy and residency requirements directly shape risk appetite: regulations like GDPR, HIPAA, or data-localization laws determine what training data may be used and where it can be stored, which bounds the acceptable risk for the chatbot. Option C is correct because model drift is a core AI risk factor — as data distributions shift, output accuracy degrades, so the committee must weigh the likelihood of drift against the organization's tolerance for degraded performance. Option D is correct because hallucination is a generative-AI-specific risk whose potential impact on brand reputation and customer trust must be assessed against the organization's risk tolerance. Option A is not a risk-appetite factor; sprint cadence is a delivery-process metric, not a measure of AI risk exposure. Option E is also not a risk-tolerance factor; GPU infrastructure cost is a budgeting concern, not a determinant of acceptable risk.

Answer

The speed of the development team's sprint cycles

Answer

The total cost of GPU infrastructure

Question 6

Which TWO metrics are most appropriate for measuring the effectiveness of an AI Risk Program?

Accepted Answer

Total number of AI-related incidents reported and remediated.. Option D is correct because tracking the total number of AI-related incidents reported and remediated directly measures how well the AI Risk Program detects, responds to, and resolves real risk events, which is a core effectiveness indicator. Option E is correct because the percentage of AI models that have undergone a formal risk assessment measures the program's coverage and consistency in applying risk controls across the AI model inventory, reflecting proactive risk management maturity. Options A, B, and C are not appropriate because average data scientist salary, coffee machine count, and social media followers are unrelated to AI risk identification, assessment, mitigation, or governance outcomes and do not indicate program effectiveness.

Answer

The average salary of data scientists.

Answer

Number of coffee machines in the building.

Answer

Number of social media followers of the company.

Question 7

Which THREE strategies are effective for mitigating 'Adversarial Risk' in an AI lifecycle?

Accepted Answer

Rate limiting on API calls to prevent brute-force querying.. Option B is correct because rate limiting on API calls throttles the query volume an attacker can send, which directly counters brute-force and model-extraction attacks that rely on high-volume probing of the inference endpoint. Option D is correct because adversarial training explicitly augments the training set with perturbed examples, hardening the model's decision boundaries against evasion and adversarial-input attacks. Option E is correct because input sanitization and validation on all model endpoints rejects malformed, out-of-distribution, or injection-style payloads before they reach the model, reducing the attack surface for adversarial manipulation. Option A does not belong because adding documentation staff addresses process and knowledge management, not adversarial manipulation of the model. Option C does not belong because a more expensive cloud provider changes cost, not the security posture against adversarial inputs.

Answer

Hiring more documentation staff.

Answer

Using a more expensive cloud provider.

Question 8

Which THREE factors should be evaluated when reviewing an AI model for 'Fairness'?

Accepted Answer

Performance disparity across different demographic cohorts.. Option A is correct because fairness review requires measuring performance disparity across different demographic cohorts — comparing metrics such as accuracy, precision, recall, or false positive/negative rates between protected groups reveals disparate impact. Option C is correct because fairness cannot be assessed without defined metrics for success that explicitly account for equity, such as equalized odds, demographic parity, or equal opportunity, which operationalize what 'fair' means for the use case. Option D is correct because training data representation across protected groups directly affects model behavior; underrepresentation or historical bias in the data propagates into biased predictions, so representation must be audited. Option B is not relevant because server uptime is an infrastructure/availability metric unrelated to model fairness. Option E is not relevant because model file size in kilobytes is a storage or deployment concern, not a fairness indicator.

Answer

The number of hours the server was powered on.

Answer

The model's file size in kilobytes.

Question 9

Which THREE types of 'Bias' need to be managed in an AI project?

Accepted Answer

Measurement bias (data collection). Measurement bias (A) must be managed because it arises during data collection when the instruments, sensors, or labeling procedures systematically distort the values recorded, so the training data does not faithfully represent the real-world phenomenon. Sampling bias (D) must be managed because it occurs during data selection when the dataset is drawn from a non-representative subset of the population, causing the model to learn patterns that do not generalize to the intended deployment context. Algorithmic bias (E) must be managed because it emerges from the model's logic, objective function, or optimization process, which can amplify or encode unfair correlations even when the input data is otherwise sound. The remaining options do not belong: the bias towards free coffee (B) is a humorous personal preference with no bearing on AI data or model behavior, and the bias of the local weather (C) is an environmental condition rather than a systematic distortion in an AI pipeline.

Answer

The bias towards free coffee

Answer

The bias of the local weather

Question 10

Which TWO strategies are recommended for 'Mitigating' AI model bias?

Accepted Answer

Applying fairness constraints during model training. Option C is correct because applying fairness constraints during model training directly mitigates bias by incorporating fairness metrics (such as demographic parity or equalized odds) into the objective function or as regularization terms, forcing the model to reduce disparate impact across protected groups. Option E is correct because training on more diverse and representative datasets addresses bias at its source—the data—by ensuring the model learns patterns from a broader population, reducing sampling bias and underrepresentation that lead to skewed predictions. Options A, B, and D are not recommended mitigation strategies: deleting all data destroys the training set and makes learning impossible, asking users to manually fix bias shifts responsibility away from systematic model design and is not scalable or reliable, and ignoring bias allows it to persist and potentially worsen in production.

Answer

Removing all data from the database

Answer

Asking the user to manually fix the bias

Answer

Ignoring the bias and hoping it goes away

Question 11

Which THREE factors are essential to consider when determining the 'AI Risk Appetite' for an enterprise?

Accepted Answer

The organization's tolerance for legal and reputational damage.. Option B is correct because AI risk appetite must be calibrated to how much legal exposure and reputational harm the organization is willing to absorb, since these are the primary consequence categories of AI failures and directly bound the acceptable risk threshold. Option C is correct because the complexity and degree of autonomy of the deployed AI models determine the severity and likelihood of adverse outcomes, so higher autonomy or opacity requires a more conservative appetite. Option D is correct because the applicable regulatory landscape and sector-specific compliance obligations (for example, EU AI Act risk tiers, HIPAA, or financial-services rules) set non-negotiable limits that the risk appetite cannot exceed. Option A is not essential because market share is a business-performance metric that does not define risk tolerance boundaries. Option E is not essential because historical data volume is an input to model development and data-governance decisions, not a determinant of how much AI risk the enterprise is willing to accept.

Answer

The current market share of the company.

Answer

Historical data volume.

Question 12

Which TWO of the following techniques help mitigate bias in the AI lifecycle?

Accepted Answer

Conducting fairness audits using specialized metrics (e.g., Disparate Impact).. Option C is correct because conducting fairness audits with specialized metrics such as Disparate Impact (the ratio of favorable outcomes for a protected group to those of a reference group, ideally near 1.0 under the 80% rule) quantitatively detects and helps correct disparate treatment or outcomes across groups in the AI lifecycle. Option D is correct because sourcing training data from diverse, representative populations reduces sampling and representation bias, ensuring the model learns patterns that generalize fairly rather than overfitting to a dominant or historically skewed subgroup. Option A is incorrect because the programming language (e.g., Java) is irrelevant to bias; bias arises from data, objectives, and evaluation, not from the implementation language. Option B is incorrect because reducing model monitoring would hide drift and emerging bias, worsening rather than mitigating it. Option E is incorrect because simply adding more features can amplify bias through proxy variables and increase overfitting without addressing underlying data or fairness issues.

Answer

Switching the programming language to Java.

Answer

Reducing the frequency of model monitoring.

Answer

Increasing the number of features in the model.

Question 13

You are integrating AI governance into the existing Corporate Governance framework. Which THREE of the following activities are essential to ensure the Board of Directors maintains adequate oversight?

Accepted Answer

Reviewing and approving the organization's AI Risk Appetite Statement.. Option A is correct because the Board's core governance role includes setting and approving the organization's risk appetite, and an AI Risk Appetite Statement translates that mandate into explicit tolerance thresholds for AI-specific risks (e.g., bias, model drift, data privacy), giving the Board a formal basis for oversight. Option C is correct because designating executive-level accountability (e.g., a Chief AI Officer or an AI risk owner at the C-suite) establishes clear ownership and escalation paths, which is a fundamental governance control ensuring the Board can hold management responsible for AI risk outcomes. Option E is correct because a defined reporting cadence for significant AI-related risk incidents and model performance trends provides the Board with timely, recurring visibility into AI risk posture, enabling informed oversight and intervention consistent with corporate governance principles. Options B and D are not Board-level governance activities: patching GPU servers is an operational IT maintenance task, and daily code reviews of training algorithms are tactical engineering controls, both of which belong to management and technical teams rather than the Board's oversight responsibilities.

Answer

Installing software patches on all GPU servers.

Answer

Conducting daily code reviews of all AI training algorithms.

Question 14

Which THREE factors contribute to 'Data Quality Risk' during the training phase?

Accepted Answer

Incomplete datasets with many missing values.. Data Quality Risk during training arises from defects in the data itself, and option A is correct because incomplete datasets with many missing values degrade the statistical representativeness of the training sample, causing the model to learn from biased or insufficient signal. Option B is correct because systematic errors in target label annotations inject consistent noise into the ground truth, so the model optimizes toward incorrect labels and produces unreliable predictions. Option D is correct because selection bias in the training set creation process means the sampled data does not reflect the real-world distribution the model will encounter, skewing learned patterns. Option C is not a data quality issue but a hardware/resource constraint (GPU memory capacity), and option E is an infrastructure/vendor choice unrelated to the intrinsic quality of the training data.

Answer

The model is too large for the GPU memory.

Answer

Using the wrong cloud vendor.

Question 15

Which THREE factors should be monitored to detect 'Model Drift' in the deployment phase?

Accepted Answer

The distribution of prediction outputs.. Model drift detection in the deployment phase focuses on changes in data and model behavior over time. Option A, the distribution of prediction outputs, is correct because a shift in the distribution of predictions (e.g., a classifier suddenly predicting one class far more often) is a direct symptom of model drift, often caused by changing input data or degraded decision boundaries. Option D, the confidence score of the model predictions, is correct because declining or abnormally shifting confidence scores indicate the model is becoming less certain and less aligned with the data it was trained on, a common signal of drift. Option E, the distribution of incoming input features, is correct because monitoring feature distributions (e.g., via statistical tests like PSI or KL divergence) detects data drift, which is the primary driver of model drift in production. Option B, the temperature of the server room, is incorrect because it is a physical infrastructure concern unrelated to statistical model behavior. Option C, the current market price of GPU hardware, is incorrect because it is a procurement/cost factor with no bearing on whether a deployed model's performance is degrading.

Answer

The temperature of the server room.

Answer

The current market price of GPU hardware.

Question 16

Which THREE items should be included in an AI 'Risk Assessment' report for a new project?

Accepted Answer

Identification of potential failure modes and their impact.. A complete report covers the risk identification, the impact analysis, and the control strategy.

Answer

The favorite color of the lead developer.

Answer

A list of all competitors currently using the same software.

Question 17

When integrating AI risk into existing ERM, which THREE aspects of an AI model lifecycle must be audited to ensure compliance?

Accepted Answer

Model validation results. Model validation results (C) must be audited because they document whether the AI model meets its stated performance, fairness, and robustness criteria before and after deployment, which is a core control for AI risk within ERM. Incident response logs (D) are essential because they provide evidence that AI-related failures, bias events, or security incidents were detected, escalated, and remediated according to the organization's risk response procedures. Data lineage and provenance (E) must be audited to verify that training, validation, and production data sources are authorized, traceable, and compliant with privacy and quality requirements, which directly affects model reliability and regulatory compliance. Personnel physical security badges (A) and office climate control settings (B) are facility and environmental controls unrelated to the AI model lifecycle, so they do not belong in an audit of AI risk integration into ERM.

Answer

Personnel physical security badges

Answer

Office climate control settings

Question 18

When designing an AI Risk Reporting dashboard for senior management, which TWO of the following should be visualized?

Accepted Answer

Status of compliance certifications for high-risk models.. Option C is correct because senior management needs visibility into the governance posture of high-risk AI models, and tracking the status of compliance certifications (e.g., ISO/IEC 42001, SOC 2, or EU AI Act conformity assessments) directly supports regulatory oversight and accountability. Option D is correct because an aggregate risk score per business unit condenses model-level risk metrics into a decision-ready KPI that lets executives compare exposure, prioritize mitigation funding, and track risk appetite across the organization. Option A is not appropriate because server uptime is an infrastructure/availability metric that belongs in IT operations dashboards, not an AI risk reporting view for executives. Option B is not appropriate because raw per-transaction prediction scores are far too granular and technical for senior management, and would create noise rather than actionable risk insight. Option E is not appropriate because employee training attendance is a tactical HR/compliance activity metric, not a strategic AI risk indicator for the board or executive committee.

Answer

Server uptime percentages.

Answer

Raw model prediction scores for every transaction.

Answer

Employee training attendance logs.

Question 19

Which THREE components are necessary for effective cross-functional AI risk coordination?

Accepted Answer

Clearly defined RACI matrix for AI initiatives. A clearly defined RACI matrix for AI initiatives (A) is necessary because it assigns explicit Responsible, Accountable, Consulted, and Informed roles across functions, eliminating ambiguity about who owns each AI risk decision and escalation path. A standardized AI risk taxonomy (C) is required so that legal, security, data science, and business teams classify and compare risks using the same categories, severity scales, and terminology, enabling consistent cross-functional assessment. A centralized AI risk register (D) is essential as the single source of truth that aggregates identified risks, owners, mitigations, and status, giving all functions shared visibility and traceability. Weekly team lunches (B) are a social nicety that may improve rapport but are not a structural mechanism for risk coordination, and unrestricted access to all production servers for all staff (E) violates least-privilege and actually increases risk rather than enabling effective coordination.

Answer

Weekly team lunches

Answer

Unrestricted access to all production servers for all staff

Question 20

Which THREE considerations must be addressed when designing a 'Training' pipeline to minimize data quality risk?

Accepted Answer

Recording full data lineage for every training run.. Option B is correct because recording full data lineage for every training run provides traceability of the exact datasets, transformations, and versions used, which is essential for detecting and remediating data quality issues and reproducing results. Option D is correct because automated outlier detection identifies noisy, anomalous, or corrupted records before they degrade model training, directly reducing data quality risk. Option E is correct because strict schema validation catches data type errors, missing fields, and format mismatches at ingestion time, preventing malformed data from entering the training pipeline. Options A and C are not correct because a high-end monitor and a specific server hardware brand are irrelevant to data quality; they concern display and procurement preferences rather than the integrity, validity, or traceability of training data.

Answer

Ensuring the training server has a high-end monitor.

Answer

Using a specific brand of server hardware.