IAPP AIGP Practice Test 74
Question 730
A vision model identifies road signs in clean weather but is highly sensitive to glare and deliberate sticker patterns. Model assurance reviewers need a documented control before approving continued experimentation. Which specific implementation best addresses the identified gap?
- Use temporal or group-aware splits and isolate evaluation records from training and tuning.
- Define and refine labeling policy with expert adjudication and representative agreement testing.
- Add robustness tests, input controls and fallback behavior for unsafe predictions.
- Validate explanation methods and document what attributions do and do not establish.
Explanation
Correct Answer
C:Add robustness tests, input controls and fallback behavior for unsafe predictions. Correct. It directly reduces or governs the specific documented risk. Security and robustness tests must cover plausible operating and adversarial conditions.
Incorrect Answers
A:Use temporal or group-aware splits and isolate evaluation records from training and tuning. Not selected. That action treats another control objective, not the specific gap described. It belongs with Dataset leakage and temporal splits: A contaminated test set can give convincing but invalid evidence of generalization. The present case instead concerns Adversarial robustness testing.
B:Define and refine labeling policy with expert adjudication and representative agreement testing. Not selected. That action treats another control objective, not the specific gap described. It belongs with Label quality and inter-rater agreement: Ground-truth quality shapes model behavior and can hide systematic harm. The present case instead concerns Adversarial robustness testing.
D:Validate explanation methods and document what attributions do and do not establish. Not selected. That action treats another control objective, not the specific gap described. It belongs with Interpretability testing: An explanation tool may be misleading unless its reliability and limitations are evaluated. The present case instead concerns Adversarial robustness testing.
Question 731
A vision model identifies road signs in clean weather but is highly sensitive to glare and deliberate sticker patterns. Model assurance reviewers need a documented control before approving continued experimentation. Which acceptance test most directly verifies the relevant control?
- Measure hazard-specific failure rates under glare, obstruction and attack conditions.
- Recreate selected features from source data and confirm sensitive transformations are visible.
- Evaluate both error differences and selection effects on representative groups after mitigation.
- Run a held-out dialect evaluation and compare false-negative rates and uncertainty thresholds.
Explanation
Correct Answer
A:Measure hazard-specific failure rates under glare, obstruction and attack conditions. Correct. It yields measurable evidence that the case-specific safeguard works. Security and robustness tests must cover plausible operating and adversarial conditions.
Incorrect Answers
B:Recreate selected features from source data and confirm sensitive transformations are visible. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Data lineage for transformations: Lineage tracks how raw data becomes model-ready features, including indirect sensitive signals. The present case instead concerns Adversarial robustness testing.
C:Evaluate both error differences and selection effects on representative groups after mitigation. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Bias measurement and mitigation: Historic training labels can reproduce inequity even when protected fields are omitted. The present case instead concerns Adversarial robustness testing.
D:Run a held-out dialect evaluation and compare false-negative rates and uncertainty thresholds. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Data quality and representativeness: Training data quality includes representativeness for the intended use population. The present case instead concerns Adversarial robustness testing.
Question 732
A vision model identifies road signs in clean weather but is highly sensitive to glare and deliberate sticker patterns. Model assurance reviewers need a documented control before approving continued experimentation. For a defensible decision, choose TWO: one diagnosis and one controlled response.
- Record end-to-end feature lineage and review proxy risks in engineered attributes.
- Select foreseeable corruption, adversarial perturbation and distribution-shift scenarios to test.
- Measure sample coverage, label consistency, subgroup representation and production-population mismatch.
- Add robustness tests, input controls and fallback behavior for unsafe predictions.
Explanation
Correct Answers
B:Select foreseeable corruption, adversarial perturbation and distribution-shift scenarios to test. Correct. As the first, evidence-gathering half of the requested pair, this establishes the specific risk boundary. Security and robustness tests must cover plausible operating and adversarial conditions.
D:Add robustness tests, input controls and fallback behavior for unsafe predictions. Correct. As the second, implementation half of the requested pair, this directly applies the needed safeguard. Security and robustness tests must cover plausible operating and adversarial conditions.
Incorrect Answers
A:Record end-to-end feature lineage and review proxy risks in engineered attributes. Not selected. It addresses a different governance mechanism, not the required assessment-and-control pair. It belongs with Data lineage for transformations: Lineage tracks how raw data becomes model-ready features, including indirect sensitive signals. The present case instead concerns Adversarial robustness testing.
C:Measure sample coverage, label consistency, subgroup representation and production-population mismatch. Not selected. It addresses a different governance mechanism, not the required assessment-and-control pair. It belongs with Data quality and representativeness: Training data quality includes representativeness for the intended use population. The present case instead concerns Adversarial robustness testing.
Question 733
A vision model identifies road signs in clean weather but is highly sensitive to glare and deliberate sticker patterns. A material training or architecture change makes the previous technical approval insufficient. Which diagnostic task establishes the risk relevant to this described system?
- Identify how sample selection, labels and historical practices may propagate protected-group disadvantages.
- Trace raw data, feature engineering, joins and transformations to the final training inputs.
- Select foreseeable corruption, adversarial perturbation and distribution-shift scenarios to test.
- Assess explanation stability, fidelity and whether the method matches model and use-case limitations.
Explanation
Correct Answer
C:Select foreseeable corruption, adversarial perturbation and distribution-shift scenarios to test. Correct. It establishes the relevant factual risk boundary before implementation. Security and robustness tests must cover plausible operating and adversarial conditions.
Incorrect Answers
A:Identify how sample selection, labels and historical practices may propagate protected-group disadvantages. Not selected. That check primarily diagnoses a different governance issue. It belongs with Bias measurement and mitigation: Historic training labels can reproduce inequity even when protected fields are omitted. The present case instead concerns Adversarial robustness testing.
B:Trace raw data, feature engineering, joins and transformations to the final training inputs. Not selected. That check primarily diagnoses a different governance issue. It belongs with Data lineage for transformations: Lineage tracks how raw data becomes model-ready features, including indirect sensitive signals. The present case instead concerns Adversarial robustness testing.
D:Assess explanation stability, fidelity and whether the method matches model and use-case limitations. Not selected. That check primarily diagnoses a different governance issue. It belongs with Interpretability testing: An explanation tool may be misleading unless its reliability and limitations are evaluated. The present case instead concerns Adversarial robustness testing.
Question 734
A vision model identifies road signs in clean weather but is highly sensitive to glare and deliberate sticker patterns. A material training or architecture change makes the previous technical approval insufficient. What is the strongest proportionate corrective design decision?
- Add robustness tests, input controls and fallback behavior for unsafe predictions.
- Apply legally appropriate sampling, fairness evaluation and mitigation with job-related validation.
- Record end-to-end feature lineage and review proxy risks in engineered attributes.
- Accept only traceable and lawfully usable data with documented source and processing stages.
Explanation
Correct Answer
A:Add robustness tests, input controls and fallback behavior for unsafe predictions. Correct. It directly reduces or governs the specific documented risk. Security and robustness tests must cover plausible operating and adversarial conditions.
Incorrect Answers
B:Apply legally appropriate sampling, fairness evaluation and mitigation with job-related validation. Not selected. That action treats another control objective, not the specific gap described. It belongs with Bias measurement and mitigation: Historic training labels can reproduce inequity even when protected fields are omitted. The present case instead concerns Adversarial robustness testing.
C:Record end-to-end feature lineage and review proxy risks in engineered attributes. Not selected. That action treats another control objective, not the specific gap described. It belongs with Data lineage for transformations: Lineage tracks how raw data becomes model-ready features, including indirect sensitive signals. The present case instead concerns Adversarial robustness testing.
D:Accept only traceable and lawfully usable data with documented source and processing stages. Not selected. That action treats another control objective, not the specific gap described. It belongs with Training data provenance: Data provenance establishes what entered training and under what conditions. The present case instead concerns Adversarial robustness testing.
Question 735
A vision model identifies road signs in clean weather but is highly sensitive to glare and deliberate sticker patterns. A material training or architecture change makes the previous technical approval insufficient. Which observable evidence best demonstrates the chosen safeguard is working?
- Run a held-out dialect evaluation and compare false-negative rates and uncertainty thresholds.
- Test for leakage indicators and compare results on a genuinely unseen time period.
- Measure hazard-specific failure rates under glare, obstruction and attack conditions.
- Re-run an approved benchmark and compare results within documented tolerances.
Explanation
Correct Answer
C:Measure hazard-specific failure rates under glare, obstruction and attack conditions. Correct. It yields measurable evidence that the case-specific safeguard works. Security and robustness tests must cover plausible operating and adversarial conditions.
Incorrect Answers
A:Run a held-out dialect evaluation and compare false-negative rates and uncertainty thresholds. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Data quality and representativeness: Training data quality includes representativeness for the intended use population. The present case instead concerns Adversarial robustness testing.
B:Test for leakage indicators and compare results on a genuinely unseen time period. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Dataset leakage and temporal splits: A contaminated test set can give convincing but invalid evidence of generalization. The present case instead concerns Adversarial robustness testing.
D:Re-run an approved benchmark and compare results within documented tolerances. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Testing evidence and reproducibility: Reproducibility makes AI testing evidence reviewable rather than anecdotal. The present case instead concerns Adversarial robustness testing.
Question 736
A vision model identifies road signs in clean weather but is highly sensitive to glare and deliberate sticker patterns. A material training or architecture change makes the previous technical approval insufficient. A material exception remains unresolved. Which containment decision best protects affected parties?
- Provide accessible non-AI service paths while underrepresented speech performance is corrected.
- Suspend adverse automated decisions when discriminatory outcomes cannot be justified or corrected.
- Withdraw unverifiable performance claims and rerun controlled tests before launch.
- Switch to a validated safe mode when input quality falls outside the assessed operating domain.
Explanation
Correct Answer
D:Switch to a validated safe mode when input quality falls outside the assessed operating domain. Correct. It limits further reliance or exposure while the risk remains unresolved. Security and robustness tests must cover plausible operating and adversarial conditions.
Incorrect Answers
A:Provide accessible non-AI service paths while underrepresented speech performance is corrected. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Data quality and representativeness: Training data quality includes representativeness for the intended use population. The present case instead concerns Adversarial robustness testing.
B:Suspend adverse automated decisions when discriminatory outcomes cannot be justified or corrected. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Bias measurement and mitigation: Historic training labels can reproduce inequity even when protected fields are omitted. The present case instead concerns Adversarial robustness testing.
C:Withdraw unverifiable performance claims and rerun controlled tests before launch. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Testing evidence and reproducibility: Reproducibility makes AI testing evidence reviewable rather than anecdotal. The present case instead concerns Adversarial robustness testing.
Question 737
An insurance model provides a feature attribution chart but explanations change wildly across near-identical cases. The development team must show evidence before promoting the next model version. Which diagnostic task establishes the risk relevant to this described system?
- Measure sample coverage, label consistency, subgroup representation and production-population mismatch.
- Trace raw data, feature engineering, joins and transformations to the final training inputs.
- Identify missing code, data, parameter, environment and test-set versions.
- Assess explanation stability, fidelity and whether the method matches model and use-case limitations.
Explanation
Correct Answer
D:Assess explanation stability, fidelity and whether the method matches model and use-case limitations. Correct. It establishes the relevant factual risk boundary before implementation. An explanation tool may be misleading unless its reliability and limitations are evaluated.
Incorrect Answers
A:Measure sample coverage, label consistency, subgroup representation and production-population mismatch. Not selected. That check primarily diagnoses a different governance issue. It belongs with Data quality and representativeness: Training data quality includes representativeness for the intended use population. The present case instead concerns Interpretability testing.
B:Trace raw data, feature engineering, joins and transformations to the final training inputs. Not selected. That check primarily diagnoses a different governance issue. It belongs with Data lineage for transformations: Lineage tracks how raw data becomes model-ready features, including indirect sensitive signals. The present case instead concerns Interpretability testing.
C:Identify missing code, data, parameter, environment and test-set versions. Not selected. That check primarily diagnoses a different governance issue. It belongs with Testing evidence and reproducibility: Reproducibility makes AI testing evidence reviewable rather than anecdotal. The present case instead concerns Interpretability testing.
Question 738
An insurance model provides a feature attribution chart but explanations change wildly across near-identical cases. The development team must show evidence before promoting the next model version. What is the strongest proportionate corrective design decision?
- Define and refine labeling policy with expert adjudication and representative agreement testing.
- Record end-to-end feature lineage and review proxy risks in engineered attributes.
- Validate explanation methods and document what attributions do and do not establish.
- Maintain reproducible experiments and protected evaluation artifacts with traceable approvals.
Explanation
Correct Answer
C:Validate explanation methods and document what attributions do and do not establish. Correct. It directly reduces or governs the specific documented risk. An explanation tool may be misleading unless its reliability and limitations are evaluated.
Incorrect Answers
A:Define and refine labeling policy with expert adjudication and representative agreement testing. Not selected. That action treats another control objective, not the specific gap described. It belongs with Label quality and inter-rater agreement: Ground-truth quality shapes model behavior and can hide systematic harm. The present case instead concerns Interpretability testing.
B:Record end-to-end feature lineage and review proxy risks in engineered attributes. Not selected. That action treats another control objective, not the specific gap described. It belongs with Data lineage for transformations: Lineage tracks how raw data becomes model-ready features, including indirect sensitive signals. The present case instead concerns Interpretability testing.
D:Maintain reproducible experiments and protected evaluation artifacts with traceable approvals. Not selected. That action treats another control objective, not the specific gap described. It belongs with Testing evidence and reproducibility: Reproducibility makes AI testing evidence reviewable rather than anecdotal. The present case instead concerns Interpretability testing.
Question 739
An insurance model provides a feature attribution chart but explanations change wildly across near-identical cases. The development team must show evidence before promoting the next model version. Which observable evidence best demonstrates the chosen safeguard is working?
- Re-run an approved benchmark and compare results within documented tolerances.
- Test similar inputs for explanation consistency and verify reviewers do not treat correlates as proven causes.
- Test for leakage indicators and compare results on a genuinely unseen time period.
- Run a held-out dialect evaluation and compare false-negative rates and uncertainty thresholds.
Explanation
Correct Answer
B:Test similar inputs for explanation consistency and verify reviewers do not treat correlates as proven causes. Correct. It yields measurable evidence that the case-specific safeguard works. An explanation tool may be misleading unless its reliability and limitations are evaluated.
Incorrect Answers
A:Re-run an approved benchmark and compare results within documented tolerances. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Testing evidence and reproducibility: Reproducibility makes AI testing evidence reviewable rather than anecdotal. The present case instead concerns Interpretability testing.
C:Test for leakage indicators and compare results on a genuinely unseen time period. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Dataset leakage and temporal splits: A contaminated test set can give convincing but invalid evidence of generalization. The present case instead concerns Interpretability testing.
D:Run a held-out dialect evaluation and compare false-negative rates and uncertainty thresholds. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Data quality and representativeness: Training data quality includes representativeness for the intended use population. The present case instead concerns Interpretability testing.
Question 740
An insurance model provides a feature attribution chart but explanations change wildly across near-identical cases. The development team must show evidence before promoting the next model version. A material exception remains unresolved. Which containment decision best protects affected parties?
- Withdraw unverifiable performance claims and rerun controlled tests before launch.
- Block unsupported causal claims and route consequential decisions for human review.
- Switch to a validated safe mode when input quality falls outside the assessed operating domain.
- Suspend adverse automated decisions when discriminatory outcomes cannot be justified or corrected.
Explanation
Correct Answer
B:Block unsupported causal claims and route consequential decisions for human review. Correct. It limits further reliance or exposure while the risk remains unresolved. An explanation tool may be misleading unless its reliability and limitations are evaluated.
Incorrect Answers
A:Withdraw unverifiable performance claims and rerun controlled tests before launch. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Testing evidence and reproducibility: Reproducibility makes AI testing evidence reviewable rather than anecdotal. The present case instead concerns Interpretability testing.
C:Switch to a validated safe mode when input quality falls outside the assessed operating domain. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Adversarial robustness testing: Security and robustness tests must cover plausible operating and adversarial conditions. The present case instead concerns Interpretability testing.
D:Suspend adverse automated decisions when discriminatory outcomes cannot be justified or corrected. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Bias measurement and mitigation: Historic training labels can reproduce inequity even when protected fields are omitted. The present case instead concerns Interpretability testing.