CCNA Aio Ai Governance Ethics Questions — Page 2 of 2

Multi-Selectmedium

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

Select 2 answers

A.It provides a measure of model confidence in its prediction

B.It requires access to the model's internal parameters

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

D.It creates an interpretable surrogate model locally around a prediction

E.It can be used with any machine learning model

AnswersD, E

LIME generates a simple model that approximates the black-box model near the instance being explained.

Why this answer

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.

Practice this question →

MCQhard

An AI governance team is implementing the NIST AI Risk Management Framework. They have identified a high-risk AI system and are in the 'Measure' function. Which activity is most appropriate for this function?

A.Conduct bias and fairness impact assessments on the model

B.Implement technical controls to mitigate identified risks

C.Document the system's intended purpose and data sources

D.Establish an AI ethics board to oversee risk decisions

AnswerA

Measuring risks includes testing for bias and fairness, which directly aligns with the Measure function.

Why this answer

In the NIST AI Risk Management Framework (AI RMF), the 'Measure' function focuses on assessing and analyzing risks associated with AI systems. For a high-risk AI system, conducting bias and fairness impact assessments is a core activity within this function, as it quantifies and evaluates potential harms related to fairness, accuracy, and transparency. This aligns with the framework's emphasis on quantitative and qualitative risk measurement before moving to risk treatment in the 'Manage' function.

Exam trap

Cisco often tests the distinction between the NIST AI RMF functions (Map, Measure, Manage, Govern) by presenting risk mitigation actions (like implementing controls) as plausible activities for the 'Measure' function, when they actually belong to the 'Manage' function.

How to eliminate wrong answers

Option B is wrong because implementing technical controls to mitigate risks belongs to the 'Manage' function, which involves risk response and treatment, not the 'Measure' function that focuses on assessment and analysis. Option C is wrong because documenting the system's intended purpose and data sources is part of the 'Map' function, which establishes context and identifies risks, not the 'Measure' function that evaluates those risks. Option D is wrong because establishing an AI ethics board is a governance structure typically associated with the 'Govern' function, which sets policies and oversight, not the 'Measure' function's risk assessment activities.

Practice this question →

MCQeasy

A company uses an AI system to generate marketing images. They are concerned about copyright ownership of the generated content. According to current US copyright law, who typically owns the copyright for AI-generated work?

A.No one; the work may be in the public domain

B.The user who provided the input prompts

C.The AI system itself, as the creator

D.The company that owns the AI model

AnswerA

According to US Copyright Office guidance, AI-generated works without substantial human authorship are not copyrightable and may be considered public domain.

Why this answer

Under US copyright law, copyright protection requires human authorship. Works created entirely by AI without human creative input are generally not copyrightable. The user who provides prompts may not own the copyright if the work lacks human authorship.

The AI developer or AI system itself cannot hold copyright.

Practice this question →

MCQmedium

A bank uses an AI system for credit scoring. To meet fairness requirements, they want to ensure the model predicts similar outcomes for individuals who are similar with respect to the target variable, regardless of protected attributes. Which fairness metric addresses this?

A.Demographic parity

B.Calibration

C.Individual fairness

D.Equalised odds

AnswerC

Individual fairness mandates that similar individuals be treated similarly.

Why this answer

Individual fairness requires that similar individuals (with respect to legitimate features) receive similar predictions. Demographic parity requires equal selection rates across groups. Equalized odds requires equality of true positive and false positive rates.

Calibration ensures predicted probabilities match actual outcomes across groups.

Practice this question →

MCQhard

A healthcare AI model is subject to GDPR because it processes data of EU patients. The model makes automated decisions about treatment plans. Under GDPR, patients have the right to obtain an explanation of the decision. Which approach BEST satisfies this 'right to explanation'?

A.Generate a SHAP-based explanation showing the contribution of each input feature to the decision

B.Offer the patient the option to have the decision reviewed by a human

C.State that the decision was made by an AI system and cannot be explained

D.Provide a copy of the model's source code to the patient

AnswerA

SHAP provides a local explanation that is both interpretable and meets the requirement to explain the decision's logic.

Why this answer

GDPR's right to explanation requires meaningful information about the logic involved in automated decisions. Providing a clear, understandable explanation using techniques like SHAP fulfills this. A simplified summary may be needed, but the explanation must be substantive.

Pointing to model cards or a confidence score alone is insufficient.

Practice this question →

MCQmedium

An insurance company uses a black-box deep learning model to set premiums. Regulators demand explanation for individual decisions. Which interpretability technique should the data science team apply to generate local explanations for each prediction?

A.Attention visualization

B.Model cards

C.LIME

D.SHAP values

AnswerC

LIME is designed specifically for local, interpretable explanations of any classifier.

Why this answer

LIME (Local Interpretable Model-agnostic Explanations) is the correct choice because it generates local, interpretable explanations for individual predictions by approximating the black-box model with a simpler, interpretable surrogate model around the specific instance. This directly meets the regulatory requirement for explaining why a particular premium was set for a specific customer, without needing access to the model's internal structure.

Exam trap

The trap here is that candidates often confuse SHAP and LIME as both being local explainers, but Cisco tests the nuance that LIME is model-agnostic and simpler to implement for deep learning black-boxes, while SHAP assumes a specific game-theoretic framework that may not align with the model's actual behavior.

How to eliminate wrong answers

Option A is wrong because attention visualization is specific to transformer-based models and highlights which input parts the model 'attends to', but it does not provide a faithful, model-agnostic local explanation for any black-box deep learning model, and it requires access to attention weights which may not be available. Option B is wrong because model cards are high-level documentation that describe overall model performance, intended use, and limitations across the entire dataset, not per-instance explanations for individual decisions. Option D is wrong because SHAP values, while providing local explanations based on cooperative game theory, are computationally expensive for deep learning models and can be misleading if the model is not additive or if feature independence assumptions are violated; LIME is more practical for quick, local explanations in a regulatory context.

Practice this question →

Multi-Selecthard

A company uses an AI model to screen job applicants. A disparate impact analysis reveals that the model's rejection rate for a protected group is significantly higher than for others. Which THREE actions should the company take to address this?

Select 3 answers

A.Revisit training data for historical bias and consider reweighting

B.Ignore the disparity because the model is accurate overall

C.Remove all demographic attributes from the dataset

D.Apply fairness constraints or adversarial debiasing during training

E.Consider using a different model that achieves better fairness metrics

AnswersA, D, E

Addressing data bias is a fundamental step to reduce disparate impact.

Why this answer

Option A is correct because revisiting the training data for historical bias and applying reweighting directly addresses the root cause of disparate impact. If the training data contains biased labels or skewed representation of the protected group, the model will learn and amplify those biases. Reweighting adjusts the loss function to give more importance to underrepresented or disadvantaged groups, helping to equalize error rates across groups.

Exam trap

Cisco often tests the misconception that removing protected attributes (option C) is sufficient to eliminate bias, when in reality it can hide bias and still allow proxy discrimination, making it an incomplete and sometimes counterproductive solution.

Practice this question →

MCQhard

A financial institution wants to deploy an AI system that automates loan approvals. Under the EU AI Act, this system would be classified as high-risk. Which of the following is a MANDATORY requirement for high-risk AI systems before market placement?

A.Obtaining a CE marking of conformity through self-assessment only

B.Establishing a risk management system throughout the system lifecycle

C.Acquiring liability insurance for AI-related harms

D.Publishing the training dataset in an open repository

AnswerB

A continuous risk management system is a mandatory requirement for high-risk AI systems under the EU AI Act.

Why this answer

The EU AI Act requires a conformity assessment (often including third-party involvement) for high-risk AI systems to ensure compliance with requirements such as data governance, transparency, and human oversight. The other options are either voluntary, not required, or apply to lower-risk tiers.

Practice this question →

MCQeasy

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?

A.The right to have all personal data deleted immediately

B.The right to receive a detailed mathematical explanation of the model

C.The right to demand a more favourable automated decision

D.The right to obtain human intervention on the part of the controller

AnswerD

GDPR Article 22 requires that meaningful human intervention be available for automated decisions producing legal effects.

Why this answer

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.

Practice this question →

MCQmedium

An AI team notices that their hiring model consistently selects male candidates over equally qualified female candidates. Analysis shows the training data contains past hiring decisions where men were predominantly hired. Which type of bias is the root cause?

A.Algorithmic bias

B.Confirmation bias

C.Selection bias

D.Historical bias

AnswerD

The data reflects past discriminatory hiring practices, causing the model to perpetuate that bias.

Why this answer

Historical bias is the root cause because the training data reflects past hiring decisions that systematically favored male candidates, encoding societal or organizational prejudices into the model. The model learns these historical patterns and perpetuates them, leading to discriminatory outcomes against equally qualified female candidates. This is distinct from algorithmic bias, which would arise from the model's design or optimization process itself.

Exam trap

Cisco often tests the distinction between historical bias (data-driven) and algorithmic bias (model-driven), and the trap here is that candidates may confuse the source of bias as being from the algorithm itself rather than the training data.

How to eliminate wrong answers

Option A is wrong because algorithmic bias refers to bias introduced by the algorithm's design, training process, or optimization function, not by the data itself. Option B is wrong because confirmation bias is a cognitive bias where individuals favor information that confirms their preexisting beliefs, which is not applicable to a machine learning model's training data. Option C is wrong because selection bias occurs when the data is not representative of the population due to non-random sampling, but here the data accurately reflects historical hiring decisions, which are themselves biased.

Practice this question →