Courseiva

CCNA Ethical Considerations of AI Questions

75 of 85 questions · Page 1/2 · Ethical Considerations of AI · Answers revealed

Practice these questionsDomain overviewAll questions
1
MCQhard

A company uses Einstein GPT to generate email responses. They want to automatically audit generated responses for potentially harmful or biased language before sending. Which Salesforce feature should they use?

A.Einstein Trust Layer.
B.Permission Set.
C.Einstein Analytics.
D.Data Mask.
AnswerA

Trust Layer provides content moderation and safety features.

Why this answer

Einstein Trust Layer is the correct feature because it provides a governance layer that automatically audits AI-generated content for toxicity, bias, and harmful language before the email is sent. It intercepts the output from Einstein GPT and applies content safety filters, ensuring compliance with responsible AI practices without requiring manual review.

Exam trap

Salesforce often tests the misconception that any 'Einstein' feature (like Analytics) can handle AI governance, but the Trust Layer is the only dedicated service for auditing and filtering AI outputs for safety and bias.

How to eliminate wrong answers

Option B is wrong because Permission Sets control user access and permissions to objects, fields, and features, not content auditing or AI safety checks. Option C is wrong because Einstein Analytics (now Tableau CRM) is a business intelligence and analytics platform for data visualization and insights, not a tool for auditing AI-generated text for harmful language. Option D is wrong because Data Mask is a security feature that obfuscates sensitive data in non-production environments, not a mechanism to audit or filter AI-generated content for bias or toxicity.

Practice this question →
2
MCQhard

A financial institution uses an AI system to approve loan applications. The system denies loans to applicants from certain postal codes at a higher rate. The model includes 'postal code' as a feature. Which ethical consideration is most directly violated?

A.Fairness
B.Privacy
C.Accountability
D.Transparency
AnswerA

Disparate impact based on postal code violates fairness principles.

Why this answer

The AI system's use of 'postal code' as a feature leads to disparate impact on applicants from certain areas, directly violating the ethical principle of fairness. Fairness requires that AI models do not discriminate against protected groups or perpetuate systemic biases, even if the feature itself is not a protected attribute. By denying loans at a higher rate based on postal code, the system is likely engaging in proxy discrimination, which is a core fairness violation.

Exam trap

Salesforce often tests the distinction between fairness and transparency, where candidates mistakenly choose transparency because they think the model's use of postal code is 'hidden' or not explainable, but the core violation is the discriminatory outcome, not the lack of explanation.

How to eliminate wrong answers

Option B (Privacy) is wrong because the issue is not about unauthorized access or misuse of personal data, but about biased outcomes from a legitimate feature. Option C (Accountability) is wrong because the question focuses on the ethical violation of the model's behavior, not on who is responsible for its deployment or oversight. Option D (Transparency) is wrong because the problem is not a lack of explainability or interpretability of the model's decisions, but the discriminatory impact of those decisions.

Practice this question →
3
Multi-Selecthard

Which THREE components are essential for an ethical AI governance framework within a large enterprise?

Select 3 answers
A.Establish a cross-functional AI ethics board.
B.Conduct regular ethical impact assessments.
C.Define clear accountability for AI outcomes.
D.Minimize human oversight to reduce operational costs.
E.Optimize for accuracy as the primary goal.
AnswersA, B, C

An ethics board brings diverse perspectives to guide AI development and use.

Why this answer

Options A, B, and C are correct. An ethics board provides oversight, impact assessments identify risks, and accountability ensures responsibility. Minimizing human oversight (D) contradicts governance.

Only focusing on accuracy (E) neglects other ethical dimensions.

Practice this question →
4
MCQmedium

A company uses Einstein Analytics to predict employee performance and identifies low-performing employees with high confidence. What is a potential ethical concern?

A.Invasion of employee privacy.
B.High computational cost.
C.Difficulty in interpreting the model.
D.Overfitting on historical data.
AnswerA

Predicting performance often uses personal data, raising privacy concerns.

Why this answer

Einstein Analytics uses machine learning models to analyze employee data and predict performance. Identifying low-performing employees with high confidence raises ethical concerns about invasion of privacy because the model may rely on sensitive personal data (e.g., communication patterns, work hours, or behavioral metrics) without explicit employee consent or transparency. This violates principles of data minimization and informed consent, which are core to ethical AI frameworks.

Exam trap

Salesforce often tests the distinction between ethical concerns (privacy, bias, transparency) and technical issues (cost, performance, overfitting), so the trap here is that candidates may confuse a model's high confidence with accuracy or fairness, overlooking that the ethical problem lies in the unauthorized use of personal data to make high-stakes predictions.

How to eliminate wrong answers

Option B is wrong because high computational cost is a technical or financial concern, not an ethical one; it does not address fairness, privacy, or bias. Option C is wrong because difficulty in interpreting the model (lack of explainability) is a separate ethical issue related to transparency, but the question specifically highlights 'high confidence' predictions, which implies the model is interpretable enough to be confident, so the core ethical concern here is privacy, not interpretability. Option D is wrong because overfitting on historical data is a model performance issue that could lead to inaccurate predictions, but it is not the primary ethical concern when the model is already identifying employees with high confidence; privacy invasion is the direct ethical risk.

Practice this question →
5
MCQeasy

A company is deploying an AI-powered chatbot for customer service. The chatbot is trained on historical support tickets. Which ethical consideration is MOST important to address before deployment?

A.Minimizing the cost of AI training
B.Ensuring the chatbot responds quickly to all queries
C.Checking for biased or discriminatory patterns in training data
D.Planning for regular model retraining
AnswerC

Bias in training data can lead to unfair or unethical outcomes.

Why this answer

Historical data may contain biased responses, leading to unfair treatment of customers. Option A is wrong because cost is a business consideration, not ethical. Option B is wrong while performance is important, it is secondary to fairness.

Option D is wrong because maintenance is operational.

Practice this question →
6
MCQhard

A company receives a complaint that their Einstein Next Best Action recommendations are consistently suggesting different products based on the customer's ZIP code, leading to unequal access. What should the company do first?

A.Contact Salesforce support for a refund.
B.Increase the number of recommendations shown.
C.Disable the recommendation engine immediately.
D.Review the training data for geographic bias.
AnswerD

Data bias is a likely cause and should be examined.

Why this answer

The first step in addressing biased AI recommendations is to investigate the root cause. Geographic bias in training data is a common source of unequal outcomes in machine learning models like Einstein Next Best Action. Reviewing the data allows the company to identify and mitigate the bias before taking any other action.

Exam trap

Salesforce often tests the misconception that the immediate reaction to AI bias should be to disable the system or escalate to support, rather than following a structured troubleshooting process that starts with data review.

How to eliminate wrong answers

Option A is wrong because requesting a refund does not address the underlying bias issue and is not a technical solution. Option B is wrong because increasing the number of recommendations shown does not fix biased recommendations; it may amplify the unequal access. Option C is wrong because disabling the recommendation engine immediately is a drastic step that may disrupt business operations without first understanding the cause of the bias.

Practice this question →
7
MCQeasy

A company is designing an AI system to screen job applicants. To ensure fairness, which practice should be implemented?

A.Use only one data source for consistency
B.Maximize the model's accuracy on historical hiring decisions
C.Conduct regular fairness audits on model outcomes
D.Remove all demographic data from the training set
AnswerC

Audits help detect and address disparate impact.

Why this answer

Regular fairness audits are essential because they systematically evaluate model outcomes for bias across demographic groups, using metrics like disparate impact or equal opportunity difference. This practice aligns with responsible AI frameworks (e.g., NIST AI Risk Management Framework) and helps detect subtle biases that may emerge from proxy variables or data drift, ensuring the screening process remains equitable over time.

Exam trap

Salesforce often tests the misconception that removing demographic data (option D) is sufficient to ensure fairness, when in reality proxy variables and model behavior must be actively monitored through audits.

How to eliminate wrong answers

Option A is wrong because using only one data source increases the risk of sampling bias and reduces the model's ability to generalize, potentially amplifying existing disparities rather than ensuring fairness. Option B is wrong because maximizing accuracy on historical hiring decisions can perpetuate and even amplify past biases (e.g., gender or racial discrimination) present in the training data, leading to unfair outcomes. Option D is wrong because simply removing demographic attributes does not eliminate bias; models can still learn proxies (e.g., zip code, name, education) that correlate with protected characteristics, a phenomenon known as 'bias through proxy variables'.

Practice this question →
8
MCQhard

A credit scoring AI uses 50 features including zip code, age, and income. The model has high accuracy but denies credit disproportionately to a protected group. An audit reveals that zip code is a proxy for race. What is the best course of action?

A.Remove zip code from the feature set and retrain.
B.Replace zip code with more relevant non-discriminatory features and retrain with fairness constraints.
C.Keep zip code but add a fairness penalty to the loss function.
D.Increase transparency by publishing the model's decision criteria.
AnswerB

Targeted feature engineering and fairness constraints mitigate bias.

Why this answer

Replacing biased proxy with more relevant features can maintain accuracy while reducing discrimination. Option A is wrong because simply removing zip code may not eliminate all proxies. Option C is wrong because retraining with same data yields same bias.

Option D is wrong because transparency alone doesn't fix bias.

Practice this question →
9
Multi-Selecteasy

Which TWO actions help ensure transparency in AI systems according to Salesforce's ethical AI guidelines?

Select 2 answers
A.Limiting access to model outputs to only a few people.
B.Using complex deep learning models without explanation.
C.Automatically retraining models weekly.
D.Documenting model assumptions and limitations.
E.Providing plain-language explanations of model predictions.
AnswersD, E

Documentation is a transparency best practice.

Why this answer

Documenting model assumptions and limitations is a core transparency practice under Salesforce's ethical AI guidelines. It ensures stakeholders understand the boundaries and potential biases of the AI system, enabling informed trust and accountability.

Exam trap

Salesforce often tests the distinction between operational actions (like retraining) and ethical governance actions (like documentation), leading candidates to mistakenly select technically beneficial but ethically irrelevant options.

Practice this question →
10
Multi-Selectmedium

Which THREE factors should an AI Associate consider when evaluating a model for potential bias?

Select 3 answers
A.The complexity of the model architecture.
B.Whether features are correlated with protected attributes.
C.Disparities in model performance metrics across groups.
D.The date the model was last deployed.
E.Whether the training data is representative of all groups.
AnswersB, C, E

Correlation can lead to proxy discrimination.

Why this answer

If a feature is correlated with a protected attribute (e.g., race, gender, age), the model may inadvertently learn and perpetuate discriminatory patterns, even if the protected attribute itself is not used as an input. This is a key source of indirect or proxy bias in machine learning systems.

Exam trap

Salesforce often tests the misconception that model complexity or deployment recency are relevant to bias detection, when in fact bias is rooted in data representation and feature correlations with protected attributes.

Practice this question →
11
Multi-Selecthard

Which TWO practices are recommended when using AI for automated decision-making in hiring?

Select 2 answers
A.Use the AI model as the sole decision-maker.
B.Regularly audit the model for adverse impact.
C.Use all available data including protected attributes.
D.Incorporate human review for high-stakes decisions.
E.Ignore adverse impact if the model is accurate.
AnswersB, D

Auditing detects bias.

Why this answer

Regular auditing for adverse impact is a core ethical practice to detect and mitigate bias in AI-driven hiring systems. Audits involve statistical analysis (e.g., the four-fifths rule from the Uniform Guidelines on Employee Selection Procedures) to compare selection rates across protected groups, ensuring the model does not disproportionately disadvantage certain demographics.

Exam trap

Salesforce often tests the misconception that model accuracy alone justifies automated decisions, tempting candidates to pick 'Ignore adverse impact if the model is accurate' (Option E) without recognizing that fairness and ethical compliance are separate, non-negotiable requirements.

Practice this question →
12
MCQeasy

A financial services company deploys an AI system to approve small business loans. The system uses a deep neural network trained on historical loan data. After deployment, an internal audit reveals that the approval rate for minority-owned businesses is 15% lower than for non-minority-owned businesses with similar financial profiles. The company's AI Ethics policy requires that AI systems be fair and transparent. The data science team has access to the training data, model architecture, and feature importance scores. The company wants to understand why the disparity exists and take corrective action. Which approach should the team take first?

A.Analyze the training data to determine if there is sampling bias or labeling bias that caused the model to associate minority ownership with higher risk.
B.Apply a disparate impact analysis to quantify the adverse impact and then adjust the decision threshold.
C.Examine the model's weights and activations to identify which features contribute to the disparity.
D.Retrain the model with a fairness constraint that penalizes disparities in approval rates.
AnswerA

Bias often stems from training data; analyzing data for imbalances or incorrect labels is the first logical step.

Why this answer

The first step in diagnosing an AI fairness issue is to audit the training data for biases such as sampling bias (e.g., underrepresentation of minority-owned businesses) or labeling bias (e.g., historical loan officers unfairly labeling minority applicants as higher risk). Since the team has access to the training data, analyzing it directly addresses the root cause of the disparity before making model-level changes. This aligns with the AI Ethics policy requirement for transparency, as data bias is a common source of unfair outcomes in deep neural networks trained on historical data.

Exam trap

Salesforce often tests the principle that data bias is the most common root cause of AI fairness issues, tempting candidates to jump to model-level fixes (like threshold adjustment or fairness constraints) instead of first auditing the training data for sampling or labeling bias.

How to eliminate wrong answers

Option B is wrong because applying a disparate impact analysis and adjusting the decision threshold treats the symptom (unequal approval rates) rather than investigating the underlying cause in the data or model; it may also violate transparency requirements if the threshold adjustment is not explainable. Option C is wrong because examining model weights and activations in a deep neural network is a black-box approach that is unlikely to reveal clear, interpretable causes of disparity, especially when feature importance scores are already available and the team should first check the data. Option D is wrong because retraining with a fairness constraint is a corrective action that should be taken only after understanding the source of bias; jumping to this step without data analysis risks introducing new biases or masking the original problem.

Practice this question →
13
MCQmedium

Refer to the exhibit. A company uses an AI model for loan approvals. The error log shows a drift warning for a specific zip code, followed by a retraining failure due to insufficient data. What is the MOST ethical concern?

A.The model may produce biased outcomes for underserved groups
B.The system failed to log the error
C.The system ignored the drift warning
D.The retraining process is too slow
AnswerA

Lack of data for a group can lead to biased predictions.

Why this answer

The drift warning indicates that the model's performance has degraded for a specific zip code, likely due to changes in the underlying data distribution. When retraining fails due to insufficient data, the model cannot adapt to these changes, which can lead to biased outcomes for underserved groups in that zip code. This is the most ethical concern because it directly impacts fairness and equity in automated decision-making.

Exam trap

Salesforce often tests the distinction between ethical concerns and operational or technical issues, so candidates may mistakenly choose a performance-related option (like retraining being too slow) instead of recognizing the fairness and bias implications of a model failing to adapt to data drift for a specific population.

How to eliminate wrong answers

Option B is wrong because the error log explicitly shows a drift warning and a retraining failure, meaning the system did log the error. Option C is wrong because the system did not ignore the drift warning; it attempted retraining but failed due to insufficient data. Option D is wrong because the retraining process being too slow is a performance issue, not the primary ethical concern; the core ethical issue is the potential for biased outcomes when retraining cannot occur.

Practice this question →
14
MCQeasy

A user asks an Einstein chatbot 'What is my current account balance?' The chatbot has been trained on transactions but is not supposed to reveal account data. Which ethical principle is at risk?

A.Privacy.
B.Accountability.
C.Transparency.
D.Fairness.
AnswerA

Customer financial information is sensitive and must be protected.

Why this answer

Revealing personal financial data violates the principle of privacy.

Practice this question →
15
MCQeasy

A company wants to use Einstein Vision for product categorization. To ensure ethical use, they should:

A.Avoid using any images that contain people.
B.Test the model for bias across different demographic groups.
C.Use only high-resolution images.
D.Only use images from a single demographic.
AnswerB

Directly addresses fairness.

Why this answer

Testing for bias across demographic groups helps ensure the model treats all users fairly.

Practice this question →
16
MCQeasy

A data scientist is training a model to predict customer churn. To ensure fairness, what should the data scientist do?

A.Focus solely on model accuracy ignoring demographic groups.
B.Ensure the training data is representative of the entire customer base.
C.Remove all demographic attributes from the dataset.
D.Use only historical data without checking for bias.
AnswerB

Representative data reduces the risk of bias.

Why this answer

Ensuring the training data is representative of the entire customer base directly addresses fairness by preventing underrepresentation or overrepresentation of specific demographic groups. A representative dataset helps the model learn unbiased patterns across all segments, reducing the risk of disparate impact. This aligns with the principle of fairness in AI, where the model's predictions should not systematically disadvantage any group.

Exam trap

Salesforce often tests the misconception that simply removing sensitive attributes (like race or gender) is sufficient to ensure fairness, when in reality the model can still learn proxies for those attributes from other correlated features.

How to eliminate wrong answers

Option A is wrong because focusing solely on model accuracy while ignoring demographic groups can lead to a model that performs well overall but has high error rates for minority groups, violating fairness principles. Option C is wrong because simply removing all demographic attributes does not guarantee fairness; the model can still learn proxies for those attributes from other correlated features (e.g., zip code for race), a phenomenon known as 'redundant encoding.' Option D is wrong because using only historical data without checking for bias propagates existing societal biases present in the data, leading to discriminatory outcomes.

Practice this question →
17
MCQeasy

A company is developing a chatbot for customer service. They want to ensure the bot does not generate offensive responses. Which practice should they implement?

A.Use unsupervised learning to allow the bot to learn naturally from conversations.
B.Remove all toxicity detection to avoid false positives.
C.Train the model only on customer service transcripts without review.
D.Deploy a content filter and human-in-the-loop moderation for sensitive interactions.
AnswerD

Content filters and human oversight prevent harmful outputs.

Why this answer

Deploying a content filter combined with human-in-the-loop moderation provides both automated detection of offensive language and human oversight for ambiguous or sensitive interactions. This layered approach ensures that the chatbot can block clearly toxic outputs in real time while allowing human reviewers to handle edge cases, reducing the risk of generating offensive responses.

Exam trap

Salesforce often tests the misconception that unsupervised learning or removing safeguards is acceptable for safety, when in fact explicit content filtering and human oversight are required to prevent offensive outputs.

How to eliminate wrong answers

Option A is wrong because unsupervised learning on raw conversations would expose the model to toxic language without guidance, likely causing it to learn and reproduce offensive patterns. Option B is wrong because removing toxicity detection eliminates the primary safeguard against offensive outputs, directly contradicting the goal of preventing harmful responses. Option C is wrong because training solely on customer service transcripts without review can still include subtle biases or inappropriate language, and the model may overfit to narrow patterns without generalizing safely.

Practice this question →
18
Multi-Selectmedium

Which THREE factors should be considered when evaluating the fairness of an AI model?

Select 3 answers
A.Disparate impact ratio across groups.
B.Overall accuracy on the test set.
C.Model training time.
D.Equal opportunity difference.
E.Demographic parity in predictions.
AnswersA, D, E

Measures adverse impact ratio.

Why this answer

The disparate impact ratio measures whether an AI model's predictions disproportionately harm or benefit certain demographic groups, typically by comparing selection rates across groups. A ratio below 0.8 or above 1.25 is often considered evidence of adverse impact, making it a key quantitative fairness metric.

Exam trap

Salesforce often tests the distinction between performance metrics (like accuracy) and fairness metrics, trapping candidates who assume a high-accuracy model is automatically fair.

Practice this question →
19
MCQmedium

An AI system is used to approve loan applications. The model uses income, zip code, and credit score as features. What is a potential ethical concern?

A.Zip code may act as a proxy for race, leading to discrimination
B.The model should be a black box to avoid bias
C.Credit scores are rarely accurate
D.Income is not a reliable predictor of repayment
AnswerA

Using zip code can indirectly discriminate based on race or ethnicity.

Why this answer

Using zip code as a feature can introduce proxy discrimination. Zip codes are strongly correlated with race and socioeconomic status due to historical redlining and residential segregation. When the model learns patterns from zip code, it may inadvertently deny loans to applicants from certain racial or ethnic groups, violating fair lending laws and ethical AI principles.

Exam trap

Salesforce often tests the misconception that bias is only introduced by explicitly using protected attributes, when in fact proxy features like zip code can cause discrimination even if race or gender is not directly used.

How to eliminate wrong answers

Option B is wrong because making the model a black box does not avoid bias; in fact, black-box models obscure how decisions are made, making it harder to detect and mitigate bias. Option C is wrong because credit scores, while not perfect, are statistically validated predictors of repayment behavior and are widely used in the financial industry; the ethical concern is not about their accuracy but about how they are combined with other features. Option D is wrong because income is a strong predictor of repayment ability; the ethical issue is not its reliability but the potential for discrimination when combined with proxy features like zip code.

Practice this question →
20
Multi-Selecthard

Which THREE strategies can help mitigate bias in an AI model? (Choose three.)

Select 3 answers
A.Remove protected attributes from training data
B.Focus training on majority group data for accuracy
C.Randomize a portion of model outputs
D.Use diverse and representative training data
E.Apply fairness metrics during model evaluation
AnswersA, D, E

Removing attributes like race/gender can prevent direct discrimination.

Why this answer

Removing protected attributes (e.g., race, gender) from training data reduces the risk of the model directly learning correlations with these sensitive features. This is a common pre-processing technique to prevent direct discrimination, though it may not eliminate indirect bias if correlated proxy features remain.

Exam trap

Salesforce often tests the misconception that simply removing protected attributes or randomizing outputs is sufficient to eliminate bias, when in fact bias can persist through proxies and requires comprehensive fairness evaluation and diverse data.

Practice this question →
21
Multi-Selectmedium

Which THREE are key ethical considerations for AI according to Salesforce?

Select 3 answers
A.Accountability
B.Profitability
C.Transparency
D.Privacy
E.Speed
AnswersA, C, D

Organizations must take responsibility for AI outcomes.

Why this answer

Options A, C, and D are correct. Privacy, transparency, and accountability are foundational ethical principles for AI. Option B is wrong because profitability is a business goal, not an ethical consideration.

Option E is wrong because speed is a performance attribute.

Practice this question →
22
MCQhard

A global retail company deploys an AI-powered chatbot for customer service. The chatbot uses natural language processing to understand and respond to customer inquiries. After deployment, the company notices that the chatbot consistently provides less accurate and less helpful responses to customers from non-English-speaking regions, particularly those using dialects or slang. The company's data science team trained the model primarily on English-language customer service transcripts from the US and UK. The AI Ethics team has raised concerns about fairness and potential bias. The company wants to address this issue while maintaining overall performance and minimizing cost. Which action should the company take first?

A.Implement a fairness constraint in the model's loss function to penalize disparities across language groups.
B.Conduct a fairness audit using diverse test cases from multiple languages and dialects to quantify the disparity.
C.Disable the chatbot for non-English languages and redirect those customers to human agents.
D.Collect more training data from all regions and retrain the model from scratch.
AnswerB

An audit with diverse test cases will identify the specific gaps, allowing targeted and cost-effective improvements.

Why this answer

The first step in addressing potential bias in an AI system is to measure and quantify the disparity. Conducting a fairness audit with diverse test cases from multiple languages and dialects provides the data science team with a clear, empirical baseline of the model's performance gaps. This diagnostic step is essential before any remediation (like retraining or adding constraints) to ensure that subsequent actions are targeted and effective, avoiding wasted resources or unintended consequences.

Exam trap

Salesforce often tests the principle that measurement and diagnosis must precede intervention; the trap here is that candidates may jump to a technical fix (like a fairness constraint) or a drastic operational change (like disabling the chatbot) without first conducting the essential diagnostic step of a fairness audit.

How to eliminate wrong answers

Option A is wrong because implementing a fairness constraint in the loss function is a technical intervention that should only be applied after the specific disparities have been identified and understood; applying it blindly can degrade overall model performance or introduce new biases without addressing the root cause. Option C is wrong because disabling the chatbot for non-English languages is a reactive, non-technical workaround that reduces service availability and customer satisfaction, failing to leverage the AI's potential and contradicting the goal of maintaining overall performance. Option D is wrong because collecting more training data from all regions and retraining from scratch is a costly, time-consuming approach that should be guided by the results of a fairness audit; without first quantifying the disparity, the new data may not address the specific failure modes, and the retraining may not be necessary if the issue can be fixed with targeted fine-tuning or data augmentation.

Practice this question →
23
MCQmedium

An admin wants to use Einstein Reply Recommendations in Service Cloud. Which ethical consideration is most important to implement before enabling the feature?

A.Reduce the cost of agent training.
B.Maximize the number of recommendations.
C.Increase the speed of case resolution.
D.Ensure customer data is anonymized.
AnswerD

Anonymization protects privacy and reduces risk of PII exposure in AI outputs.

Why this answer

Customer data privacy is paramount; anonymizing data ensures recommendations do not expose sensitive information.

Practice this question →
24
Multi-Selecthard

A company is developing an AI system to assist with hiring. Which TWO practices are essential for ethical AI deployment?

Select 2 answers
A.Optimize the system for speed to reduce waiting times
B.Remove all demographic features to ensure fairness
C.Conduct regular bias audits on model predictions
D.Maximize accuracy on historical hiring data
E.Obtain informed consent from applicants if their data is used
AnswersC, E

Audits help detect and mitigate discriminatory outcomes.

Why this answer

Regular bias audits are a core ethical practice for AI systems, especially in hiring. These audits involve systematically testing the model's predictions across demographic groups to detect and mitigate unintended discrimination, ensuring compliance with fairness standards like the EEOC's Uniform Guidelines on Employee Selection Procedures.

Exam trap

Salesforce often tests the misconception that removing demographic features (Option B) is sufficient to eliminate bias, when in fact proxy variables and model behavior must be actively monitored through audits (Option C).

Practice this question →
25
MCQhard

A company's Einstein Discovery model for customer lifetime value shows a significant correlation between predicted value and customer's postal code. The company is concerned about ethical implications. What is the most appropriate response?

A.Remove the postal code field from the model immediately
B.Investigate whether postal code is a proxy for protected attributes and, if so, consider retraining the model without it or with fairness constraints
C.Add more demographic data to the model to improve its accuracy
D.Ignore the correlation since the model is predicting business value, not demographic attributes
AnswerB

This approach addresses the ethical concern while preserving model utility.

Why this answer

(Investigate whether postal code is a proxy for protected attributes and, if so, consider retraining the model without it or with fairness constraints) is correct because postal code can be a proxy for race or income. Option A (removing postal code outright) may not be straightforward. Option C (ignoring correlation as coincidental) is unethical.

Option D (adding more demographic data) could increase bias.

Practice this question →
26
Multi-Selectmedium

An AI system is used to detect fraud in financial transactions. Which THREE steps should be taken to address ethical concerns?

Select 3 answers
A.Lower the fraud detection threshold to catch more cases
B.Automatically accept all flagged transactions to improve user experience
C.Implement a human-in-the-loop for high-stakes decisions
D.Ensure the model provides explanations for its decisions
E.Test the model for disparate impact across demographic groups
AnswersC, D, E

Human oversight ensures accountability.

Why this answer

Implementing a human-in-the-loop ensures that high-stakes decisions, such as blocking a legitimate transaction or allowing a potentially fraudulent one, are reviewed by a human before final action. This addresses ethical concerns by preventing fully automated decisions that could cause financial harm or violate user trust, and it aligns with principles of accountability and fairness in AI governance.

Exam trap

Salesforce often tests the misconception that ethical AI is solely about improving model performance or user experience, when in fact it requires balancing accuracy, fairness, and human oversight—candidates may incorrectly choose options that sound beneficial (like lowering thresholds) without considering the ethical trade-offs.

Practice this question →
27
MCQmedium

Refer to the exhibit. A Salesforce admin is reviewing an AI model's fairness report. Which action should the admin take?

A.Remove the email_engagement feature to improve fairness.
B.Retrain the model because the equal opportunity score is below threshold.
C.Increase the bias threshold to 0.9.
D.Deploy the model because all metrics exceed the threshold.
AnswerB

The low equal opportunity score indicates bias that needs mitigation.

Why this answer

The equal opportunity score (0.72) is below the bias threshold (0.8), indicating potential unfairness in true positive rates across groups. Option A is wrong because demographic parity is above threshold but equal opportunity is not, so not all metrics exceed threshold. Option C is wrong because removing features may not address the root cause.

Option D is wrong because increasing the threshold would mask the problem.

Practice this question →
28
MCQmedium

Refer to the exhibit. Based on the JSON policy for AI fairness checks, which fairness metric is NOT enabled?

A.Demographic parity
B.All are enabled
C.Disparate impact
D.Equal opportunity
AnswerD

Correct. The 'equal_opportunity' field is false.

Why this answer

(Equal opportunity) is correct because the JSON policy shown in the exhibit configures fairness checks for demographic parity, disparate impact, and equalized odds, but does NOT include the equal opportunity metric. Equal opportunity requires equal true positive rates across groups, which is a separate metric from equalized odds and must be explicitly enabled in the policy definition.

Exam trap

Salesforce often tests the distinction between equalized odds and equal opportunity, trapping candidates who assume equalized odds automatically includes equal opportunity, when in fact they are separate metrics with different mathematical definitions.

How to eliminate wrong answers

Option A is wrong because demographic parity is explicitly enabled in the JSON policy under the 'fairness_metrics' array. Option B is wrong because not all metrics are enabled; the policy omits equal opportunity. Option C is wrong because disparate impact is also explicitly listed in the policy's fairness metrics.

Practice this question →
29
MCQhard

A company deploys an AI recommender system that personalizes content. The system is trained on user click data. After deployment, the company notices that the system increasingly recommends sensationalist content, leading to user polarization. Which principle is being violated?

A.Accuracy
B.Privacy
C.Beneficence
D.Transparency
AnswerC

The system should promote well-being and avoid harm.

Why this answer

The recommender system's shift toward sensationalist content, which polarizes users, violates the principle of beneficence because it causes harm (user polarization) rather than promoting well-being. Beneficence requires AI systems to act in the best interests of users and society, not to optimize for engagement metrics at the expense of ethical outcomes.

Exam trap

Salesforce often tests the distinction between ethical principles by presenting a scenario where a system functions correctly (accurate) but produces harmful outcomes, leading candidates to mistakenly choose accuracy or transparency instead of beneficence.

How to eliminate wrong answers

Option A is wrong because accuracy refers to the system's ability to make correct predictions or recommendations based on training data, not to the ethical impact of those recommendations; the system may be accurately predicting clicks on sensationalist content. Option B is wrong because privacy concerns unauthorized access or misuse of personal data, whereas the issue here is about the content being recommended, not data exposure. Option D is wrong because transparency involves explainability and openness about how the system works, but the problem is the harmful outcome of the recommendations, not a lack of clarity in the system's logic.

Practice this question →
30
MCQhard

An AI model predicts employee performance. The HR team uses it to identify high-potential employees. What is a potential ethical risk?

A.Over-reliance on the model
B.Privacy violation
C.Underutilization of human judgment
D.All of the above
AnswerD

Correct. All listed risks are potential ethical concerns.

Why this answer

All three listed risks—over-reliance on the model, privacy violation, and underutilization of human judgment—are potential ethical risks when an AI model predicts employee performance. Over-reliance can lead to automated decisions without human oversight, privacy violation may occur if sensitive employee data is mishandled, and underutilization of human judgment ignores contextual factors that the model cannot capture. Together, these represent a comprehensive set of ethical concerns in AI-driven HR practices.

Exam trap

Salesforce often tests the 'all of the above' trap where candidates think only one or two risks apply, but the question explicitly lists multiple interconnected ethical concerns that collectively form the correct answer.

How to eliminate wrong answers

Option A is wrong because over-reliance on the model is indeed a risk, but it is not the only risk, so selecting only A ignores other ethical issues. Option B is wrong because privacy violation is a valid risk, but it is incomplete without considering over-reliance and underutilization of human judgment. Option C is wrong because underutilization of human judgment is a real concern, but it does not cover the full spectrum of risks including privacy and over-reliance.

Practice this question →
31
MCQeasy

A user asks an AI assistant to generate content that may be offensive. What should the AI do?

A.Ignore the request
B.Refuse and explain why
C.Generate and report the user
D.Generate with a warning
AnswerB

Correct. The AI should not produce offensive content and should provide reasoning.

Why this answer

The AI should refuse the request and explain why, upholding ethical standards.

Practice this question →
32
MCQeasy

A Salesforce admin wants to deploy an Einstein bot that uses natural language processing. Which practice best ensures ethical use?

A.Provide clear disclaimers that the user is interacting with an AI.
B.Use the bot only for internal processes.
C.Collect as much personal data as possible to improve accuracy.
D.Allow the bot to make autonomous decisions without human review.
AnswerA

Clear disclaimers ensure transparency and informed consent.

Why this answer

Transparency is a key ethical principle; users should know they are interacting with AI. Option B is wrong because restricting to internal processes does not address ethical use. Option C is wrong because collecting excessive personal data violates privacy.

Option D is wrong because autonomous decisions may require human oversight.

Practice this question →
33
Multi-Selectmedium

Which TWO actions are essential for ensuring transparency in an AI system? (Choose two.)

Select 2 answers
A.Hide the model's internal logic to protect intellectual property
B.Log all AI decisions and allow audit
C.Train the model on the largest dataset available
D.Provide clear explanations for AI decisions
E.Obtain consent from all data subjects
AnswersB, D

Auditability is essential for transparency.

Why this answer

Options B and D are correct because providing explanations and logging decisions are key to transparency. Option A is wrong because hiding the model reduces transparency. Option C is wrong because training on all data may embed biases, and does not directly relate to transparency.

Option E is wrong because consent is about privacy, not transparency.

Practice this question →
34
MCQmedium

An AI Associate deploys an Einstein Bot that uses sentiment analysis to escalate frustrated customers. After launch, the bot escalates disproportionately for non-native English speakers. What is the most likely cause?

A.The sentiment model was trained on a non-representative dataset.
B.The bot is routing to the wrong department.
C.The escalation threshold is set too low.
D.The bot is not properly connected to the escalation queue.
AnswerA

Training data lacking linguistic diversity causes biased sentiment detection.

Why this answer

The sentiment analysis model likely exhibits bias due to training data that does not adequately represent the linguistic patterns, idioms, or expressions of non-native English speakers. This causes the model to misinterpret neutral or positive statements from these users as negative or frustrated, leading to disproportionate escalations. A non-representative dataset is a common source of algorithmic bias in AI systems.

Exam trap

Salesforce often tests the concept that bias in AI systems typically originates from the training data or model design, not from operational configuration issues like thresholds or routing, which are common distractors.

How to eliminate wrong answers

Option B is wrong because routing to the wrong department would cause misdirected escalations, not a disproportionate escalation rate for a specific demographic group. Option C is wrong because a low escalation threshold would increase escalations across all users uniformly, not selectively for non-native English speakers. Option D is wrong because a disconnected escalation queue would prevent any escalations from being processed, not cause selective over-escalation.

Practice this question →
35
MCQmedium

Refer to the exhibit. What is the most likely cause of the fairness issue?

A.The model overfits to the male group.
B.The training data is imbalanced, causing the model to perform better on the majority group.
C.The overall accuracy is too low.
D.The model is inherently biased against females.
AnswerB

Imbalanced data leads to unequal performance.

Why this answer

Imbalanced training data often leads to disparate performance. Option A is wrong because the model is not inherently biased. Option C is wrong because overall accuracy can be high despite bias.

Option D is wrong because there is no indication of overfitting.

Practice this question →
36
Multi-Selecthard

Which TWO actions best promote transparency in an AI system?

Select 2 answers
A.Limit access to the model's logic to protect intellectual property.
B.Publish an audit trail of model inputs and decisions.
C.Use a complex deep learning model for higher accuracy.
D.Provide clear explanations for individual predictions.
E.Remove feature importance to simplify the model.
AnswersB, D

Audit trails provide insight into decision process.

Why this answer

Publishing an audit trail of model inputs and decisions enables external verification of the AI system's behavior, which is a core requirement for transparency. This allows stakeholders to trace how specific inputs led to particular outputs, ensuring accountability and facilitating debugging or compliance audits.

Exam trap

Salesforce often tests the misconception that transparency is about protecting the model or maximizing accuracy, when in fact it is about openness and explainability of decisions.

Practice this question →
37
MCQeasy

A company wants to deploy an AI system that makes hiring decisions. To comply with ethical guidelines, what should they do before deployment?

A.Conduct an ethics review and perform bias testing on diverse datasets.
B.Ensure the system achieves high accuracy and ignore other metrics.
C.Deploy immediately and monitor for issues.
D.Test the system only on a small dataset to expedite launch.
AnswerA

Ethics review and bias testing are proactive measures.

Why this answer

Conducting an ethics review and performing bias testing on diverse datasets are essential steps to identify and mitigate potential discriminatory outcomes in AI-driven hiring systems. This aligns with ethical AI frameworks that require fairness, accountability, and transparency before deployment, ensuring the model does not perpetuate historical biases or violate anti-discrimination laws.

Exam trap

Salesforce often tests the misconception that high accuracy alone guarantees ethical AI, when in fact fairness metrics and bias testing are mandatory to prevent discriminatory outcomes in high-stakes applications like hiring.

How to eliminate wrong answers

Option B is wrong because prioritizing only high accuracy can mask harmful biases; a model may achieve high overall accuracy but still systematically discriminate against protected groups due to imbalanced data or proxy features. Option C is wrong because deploying immediately without prior testing violates ethical guidelines and can lead to real-world harm, legal liability, and loss of trust; monitoring alone cannot retroactively fix embedded biases. Option D is wrong because testing on a small dataset is insufficient to detect bias across diverse demographic groups and may lead to overfitting or failure to uncover edge cases, undermining the reliability and fairness of the system.

Practice this question →
38
Multi-Selecteasy

A Salesforce administrator deploys an Einstein Bot. Which TWO ethical considerations should be addressed? (Choose two.)

Select 2 answers
A.The bot should disclose it is an AI
B.The bot should mimic a human
C.The bot should make decisions autonomously
D.The bot should not collect personal data
E.The bot should escalate to a human when needed
AnswersA, E

Correct. Transparency requires the bot to identify itself as AI.

Why this answer

Ethical AI guidelines, including those from Salesforce, require that bots disclose their non-human identity to users. This transparency builds trust and ensures users are aware they are interacting with an AI, not a human, which is a core ethical principle in AI deployment.

Exam trap

Salesforce often tests the misconception that ethical AI means bots should never collect personal data, but the real ethical requirement is transparency and consent, not an absolute prohibition on data collection.

Practice this question →
39
Multi-Selectmedium

A company wants to ensure their AI model complies with ethical guidelines. Which TWO actions are essential? (Choose two.)

Select 2 answers
A.Avoid transparency
B.Provide human oversight
C.Use the most complex model
D.Automate all decisions
E.Document model decisions
AnswersB, E

Correct. Human oversight ensures decisions can be reviewed and overridden.

Why this answer

Human oversight (Option B) is essential because it ensures that AI decisions can be reviewed, overridden, or corrected by a person, which is a core requirement of ethical AI frameworks such as the EU AI Act and NIST AI Risk Management Framework. This oversight helps catch biased outputs, edge cases, or harmful actions that the model might produce, maintaining accountability and safety.

Exam trap

Salesforce often tests the misconception that 'automation' is always the goal of AI, but the trap here is that ethical guidelines require human oversight and documentation, not full automation or complexity.

Practice this question →
40
MCQmedium

An AI system used for medical diagnosis occasionally produces incorrect results. A doctor notices the errors but continues using the system without reporting them. Which ethical principle is primarily at risk?

A.Fairness
B.Transparency
C.Privacy
D.Accountability
AnswerD

Healthcare professionals are accountable for AI-assisted decisions and must report errors.

Why this answer

Accountability means humans must oversee AI decisions and report issues. Option B is wrong because transparency is about disclosure. Option C is wrong because fairness is about bias.

Option D is wrong because privacy is about data protection.

Practice this question →
41
MCQeasy

A developer is creating a custom AI model on Salesforce. To ensure the model is fair across demographic groups, which activity should be included in the development process?

A.Feature selection using correlation matrix.
B.Bias testing using a diverse test dataset.
C.Cross-validation to avoid overfitting.
D.Hyperparameter tuning with grid search.
AnswerB

This evaluates model performance across demographics.

Why this answer

Bias testing using diverse datasets directly evaluates fairness across groups.

Practice this question →
42
MCQmedium

A healthcare organization is deploying an AI model to predict patient readmission risk. The model was trained on historical data that underrepresented minority populations. During testing, the model shows lower accuracy for those groups. What should the data scientist do first?

A.Remove sensitive attributes like race and gender from the training data.
B.Ignore the disparity because the model's overall accuracy is acceptable.
C.Retrain the model with more complex algorithms to improve accuracy.
D.Re-evaluate the training data to ensure balanced representation and consider re-sampling techniques.
AnswerD

Ensuring data representativeness addresses root cause of bias.

Why this answer

The first step in addressing model bias is to audit the training data for representational imbalance. Re-evaluating the data and applying re-sampling techniques (e.g., oversampling minority groups or undersampling the majority) directly targets the root cause of the disparity—skewed class distributions—before modifying the model or its features.

Exam trap

Salesforce often tests the misconception that removing sensitive attributes or improving model complexity automatically fixes bias, when in fact the data imbalance must be addressed first at the dataset level.

How to eliminate wrong answers

Option A is wrong because simply removing sensitive attributes does not eliminate bias; proxy variables (e.g., ZIP code, income) can still encode race or gender, and the model may still learn biased correlations from remaining features. Option B is wrong because ignoring the disparity violates ethical AI principles and regulatory expectations (e.g., FDA or HIPAA guidelines for healthcare models), and overall accuracy can mask significant harm to underrepresented groups. Option C is wrong because using more complex algorithms (e.g., deeper neural networks) does not fix biased training data; it may even amplify existing disparities by overfitting to the majority class patterns.

Practice this question →
43
MCQhard

An insurance company uses an AI model to set auto insurance premiums. The model uses factors including driving history, age, and ZIP code. A regulator finds that premiums in certain low-income neighborhoods are significantly higher than in affluent neighborhoods with similar risk profiles. The company's actuaries argue that the model is actuarially sound because it accurately predicts claims based on historical data. The company wants to comply with ethical guidelines and avoid legal action. Which action should they take?

A.Defend the model based on its actuarial accuracy and historical claims data.
B.Incorporate a fairness constraint that requires similar premiums for similar risk profiles regardless of ZIP code.
C.Cap premium increases in low-income neighborhoods at a fixed percentage.
D.Remove ZIP code from the model inputs entirely.
AnswerB

This ensures fairness while preserving the model's ability to differentiate based on actual risk.

Why this answer

Introducing a fairness check ensures that similar risk levels result in similar premiums across neighborhoods, addressing ethical concerns without discarding valid risk factors. Option A ignores the issue. Option C removes a potentially relevant factor, but may reduce accuracy.

Option D is a band-aid that doesn't fix underlying bias.

Practice this question →
44
MCQhard

An AI system for hiring is deployed. After six months, the HR team notices that the model's recommendations closely mimic past human hires, which were biased. The team wants to correct this. What should be their first step?

A.Shut down the AI system entirely
B.Implement continuous monitoring and a feedback loop to detect and mitigate bias
C.Retrain the model with the same historical data but with more features
D.Make the model's decision process fully transparent to all candidates
AnswerB

Monitoring allows ongoing adjustment to ensure fairness.

Why this answer

Continuous monitoring and feedback loops can detect and correct drift or bias. Option A is wrong because removing the model does not solve underlying bias. Option C is wrong because past data already contains bias.

Option D is wrong because complete transparency does not automatically correct bias.

Practice this question →
45
MCQhard

An organization uses Einstein Discovery to analyze survey data. The model reveals a correlation between age and satisfaction. What is the responsible use of this insight?

A.Act on the correlation immediately
B.Investigate causality before action
C.Discard the result
D.Publish the result as-is
AnswerB

Correct. Investigation ensures decisions are based on sound reasoning.

Why this answer

Correlation does not imply causation; responsible use involves investigating the underlying cause before acting.

Practice this question →
46
MCQmedium

A company deploys an AI system that makes decisions about loan approvals. For transparency, what should they provide to applicants?

A.Explanation of factors considered
B.The training data
C.The full algorithm
D.Confidence scores
AnswerA

Correct. Providing the key factors used in the decision meets transparency requirements.

Why this answer

Transparency in AI-driven loan approvals requires providing applicants with an explanation of the factors considered in the decision, such as credit score, income, or debt-to-income ratio. This aligns with ethical AI principles like explainability and fairness, enabling applicants to understand and potentially contest the decision. Providing the full algorithm or training data would expose proprietary information and potentially violate data privacy regulations like GDPR.

Exam trap

Salesforce often tests the distinction between transparency (explaining the decision) and disclosure (revealing the model internals), trapping candidates who think providing the full algorithm or training data is necessary for transparency.

How to eliminate wrong answers

Option B is wrong because providing the training data would reveal sensitive personal information of other applicants, violate data privacy laws (e.g., GDPR, CCPA), and could introduce bias or security risks without helping the individual understand their specific decision. Option C is wrong because disclosing the full algorithm would expose proprietary intellectual property, enable gaming of the system, and is unnecessary for transparency—explainability focuses on decision factors, not code. Option D is wrong because confidence scores alone do not explain why a decision was made; they only indicate the model's certainty, which lacks the actionable reasoning required for transparency and regulatory compliance.

Practice this question →
47
MCQeasy

A sales team uses Einstein Lead Scoring. They notice leads from certain industries are always low-scored. What should they do?

A.Retrain the model weekly
B.Ignore the scores
C.Use a different AI system
D.Review training data for bias
AnswerD

Bias in training data can cause unfair scoring across industries.

Why this answer

Low scores for specific industries often indicate bias in the training data, where historical lead data may have underrepresented or mislabeled those industries. Reviewing the training data for bias allows the team to identify and correct such imbalances, ensuring the Einstein Lead Scoring model produces fair and accurate predictions across all segments.

Exam trap

Salesforce often tests the misconception that retraining or replacing the AI system is the solution to bias, when in fact the root cause lies in the training data, not the model or its update frequency.

How to eliminate wrong answers

Option A is wrong because retraining the model weekly does not address the root cause of bias; if the training data itself is biased, more frequent retraining will only perpetuate the same skewed patterns. Option B is wrong because ignoring the scores defeats the purpose of using AI-driven lead scoring and can lead to missed opportunities or misallocated sales efforts. Option C is wrong because switching to a different AI system does not guarantee unbiased scoring; without addressing the underlying data bias, any model trained on the same flawed data will exhibit similar issues.

Practice this question →
48
MCQhard

Refer to the exhibit. A company configures a Prompt Builder policy for Einstein GPT. What is the primary role of the 'checkPromptOutput' flag?

A.To log all prompts for audit purposes.
B.To scan the generated text against the banned words list.
C.To limit the total number of tokens in the generated response.
D.To send the output to a human reviewer before sending.
AnswerB

checkPromptOutput likely enables content scanning.

Why this answer

The 'checkPromptOutput' flag in a Prompt Builder policy for Einstein GPT is specifically designed to scan the generated text against a banned words list. This ensures that the AI output does not contain prohibited or sensitive terms, aligning with ethical and compliance requirements. It is a content filtering mechanism, not a logging, token-limiting, or human-review function.

Exam trap

The trap here is that candidates often confuse content filtering (banned words scanning) with broader safety mechanisms like logging, token limits, or human review, because all are related to output control but serve distinct purposes.

How to eliminate wrong answers

Option A is wrong because logging prompts for audit purposes is typically handled by separate audit trail or logging configurations, not the 'checkPromptOutput' flag which focuses on real-time content scanning. Option C is wrong because limiting the total number of tokens in the generated response is controlled by token limit parameters or max tokens settings, not by a flag that checks for banned words. Option D is wrong because sending output to a human reviewer before sending is a human-in-the-loop workflow, often managed by approval policies or review queues, not by the 'checkPromptOutput' flag which automates filtering without human intervention.

Practice this question →
49
MCQhard

A financial institution deploys an AI system to recommend investment portfolios to retail clients. The system uses reinforcement learning to maximize returns based on client risk profiles. After six months, an internal audit reveals that the system has been consistently recommending high-risk, high-commission products to elderly clients with low risk tolerance, resulting in significant financial losses for those clients. The system's training data included historical transactions, which showed that elderly clients were less likely to complain or switch advisors. The institution's AI ethics policy mandates fairness, transparency, and accountability. The system currently provides no explanations for its recommendations, and there is no human oversight process. The compliance team needs to remediate the situation. Which course of action BEST addresses the ethical violations?

A.Disable the AI system and revert to manual portfolio management.
B.Add a disclaimer to all recommendations stating that past performance does not guarantee future results.
C.Adjust the model to lower the risk threshold for all clients.
D.Retrain the model on a balanced dataset, implement explainability features, and require human approval for high-risk recommendations to elderly clients.
AnswerD

This addresses bias, transparency, and accountability.

Why this answer

It directly addresses the root cause of the ethical violations: biased training data (historical transactions where elderly clients were less likely to complain) and lack of transparency. Retraining on a balanced dataset mitigates the reinforcement learning model's exploitation of that bias, while explainability features (e.g., SHAP values or LIME) and human-in-the-loop approval for high-risk recommendations ensure accountability and fairness as mandated by the AI ethics policy.

Exam trap

Salesforce often tests the misconception that a single technical fix (like lowering risk thresholds or adding disclaimers) is sufficient to resolve ethical violations, when in fact a multi-pronged approach addressing data bias, transparency, and human oversight is required.

How to eliminate wrong answers

Option A is wrong because disabling the AI system and reverting to manual management is a reactive, non-scalable solution that does not address the underlying bias or provide a path to compliant AI deployment; it also ignores the potential benefits of AI when properly governed. Option B is wrong because adding a disclaimer does not fix the biased recommendations or lack of transparency; it merely shifts legal liability without correcting the model's unethical behavior or providing explanations. Option C is wrong because lowering the risk threshold for all clients is a blunt, one-size-fits-all approach that disregards individual risk profiles and may still result in inappropriate recommendations for elderly clients with low risk tolerance; it does not address the training data bias or the need for explainability and human oversight.

Practice this question →
50
MCQeasy

When implementing AI in Salesforce, which practice best supports the ethical principle of transparency?

A.Provide human-readable explanations for each AI prediction
B.Use proprietary algorithms without disclosing their logic
C.Deploy a complex neural network model without interpretability features
D.Only report overall model accuracy metrics to end users
AnswerA

Explanations enable understanding and trust.

Why this answer

Transparency requires that the logic and outcomes of AI systems are understandable. Option A is correct because providing explanations for predictions allows users to understand and trust the AI. Option B (keeping proprietary algorithms secret) hinders transparency.

Option C (using complex models without explanation) obscures decision-making. Option D (only reporting accuracy metrics) does not explain specific decisions.

Practice this question →
51
MCQeasy

A company uses an AI model to screen job applicants. They discover the model is rejecting candidates from a certain demographic at a higher rate. Which ethical principle is most clearly violated?

A.Fairness
B.Transparency
C.Privacy
D.Accountability
AnswerA

Correct. The model's bias against a demographic violates fairness.

Why this answer

Fairness requires that AI systems do not discriminate against groups. The model's disparate impact violates fairness.

Practice this question →
52
MCQhard

An AI Associate reviews the Lead Scoring model exhibit. What is the primary ethical concern with this model?

A.The model uses too many features.
B.The model has low recall, potentially missing minority class leads.
C.The model is not explainable.
D.The training data is imbalanced.
AnswerB

Low recall can lead to underrepresentation of certain groups.

Why this answer

The primary ethical concern is that the model has low recall, meaning it fails to identify a significant portion of actual positive leads (the minority class). In a lead scoring context, this can result in missed business opportunities and potential bias against certain customer segments, as the model systematically overlooks valuable leads that do not fit the majority pattern.

Exam trap

Salesforce often tests the distinction between a technical problem (like imbalanced data) and its ethical consequence (like low recall causing unfair outcomes), so candidates mistakenly pick the technical cause (D) instead of the ethical impact (B).

How to eliminate wrong answers

Option A is wrong because using many features is not inherently an ethical concern; feature selection impacts performance and overfitting, but the ethical issue here is about fairness and missed opportunities, not feature count. Option C is wrong because the exhibit does not indicate a lack of explainability; the model could be a decision tree or logistic regression that is inherently interpretable, and explainability is not the primary ethical issue raised by the confusion matrix. Option D is wrong because imbalanced training data is a technical challenge that can lead to low recall, but the primary ethical concern is the consequence of that imbalance—specifically the model's low recall causing minority class leads to be missed—not the imbalance itself.

Practice this question →
53
MCQhard

A financial services firm deployed an AI model to automate loan approvals. The model was trained on historical loan data from the past 10 years, which shows that applicants from certain zip codes have higher default rates. After six months, the company's compliance team receives complaints that applicants from predominantly low-income neighborhoods are being rejected at a much higher rate than applicants from affluent areas, even when their financial profiles are similar. The model's overall accuracy remains high (95%), and the loan default rate has decreased by 15% since deployment. The company wants to address the ethical concerns without sacrificing performance. Which course of action should the company take?

A.Remove the zip code feature from the model inputs.
B.Retrain the model with a balanced dataset that includes more examples from underrepresented neighborhoods and enforce fairness constraints.
C.Adjust the approval threshold lower only for applicants from low-income neighborhoods.
D.Continue using the existing model since it has high accuracy and reduces defaults.
AnswerB

Balanced data reduces bias and fairness constraints ensure equitable treatment, aligning with ethical AI principles.

Why this answer

Retraining with balanced data mitigates the representation bias, addressing the root cause. Option A ignores the fairness issue. Option C removes a feature that may be a proxy for other factors, but it may not eliminate bias if other correlated features remain.

Option D adjusts thresholds only for some groups, which could be considered unfair and may not be accepted by regulators.

Practice this question →
54
MCQhard

A social media platform uses an AI model to automatically detect and remove hate speech. The model uses natural language processing and was trained on public posts. Recently, an internal audit reveals that the model removes posts from minority ethnic groups at a rate 3 times higher than from majority groups, even when the content is similar. The model achieves high precision and recall on the test set. The platform's content moderation team is overwhelmed with appeals. The company wants to maintain a safe environment while being fair. Which approach best addresses both goals?

A.Disable the AI moderation and rely solely on user reports.
B.Conduct an audit of the training data to identify gaps, then retrain with more representative data including diverse examples of hate speech and non-hate speech.
C.Add more human moderators to review all flagged content from minority groups.
D.Adjust the detection threshold only for minority group posts to reduce flags.
AnswerB

This tackles the root cause of bias: underrepresentation of certain groups in training data leads to over-sensitivity.

Why this answer

A comprehensive audit and retraining with diverse data addresses the bias at the root. Option A gives special treatment that could be seen as unfair. Option C removes moderation, risking harmful content.

Option D does not solve the underlying bias.

Practice this question →
55
MCQeasy

A company is deploying an AI-powered chatbot to handle customer service inquiries. The bot uses historical chat data for training. Which ethical consideration is MOST important to address before deployment?

A.Maximizing the chatbot's response accuracy
B.Obtaining consent from customers whose data is used for training
C.Ensuring the chatbot can handle high traffic volumes
D.Designing a human handoff protocol for complex issues
AnswerB

Using customer data requires informed consent and adherence to privacy laws.

Why this answer

The most critical ethical consideration is obtaining consent from customers whose historical chat data is used to train the chatbot. Under regulations like GDPR and CCPA, personal data (including chat transcripts) requires explicit consent for processing, especially when used to train an AI model. Deploying without consent violates data privacy laws and erodes user trust, regardless of the chatbot's technical performance.

Exam trap

Salesforce often tests the distinction between ethical obligations (like consent) and technical or operational features (like accuracy or scalability), leading candidates to confuse a performance metric with a compliance requirement.

How to eliminate wrong answers

Option A is wrong because maximizing response accuracy is a performance goal, not an ethical consideration; it does not address the legal and moral requirement for data consent. Option C is wrong because ensuring high traffic handling is a scalability and infrastructure concern, unrelated to the ethical principle of data privacy and consent. Option D is wrong because designing a human handoff protocol is a user experience and operational safety measure, not the primary ethical issue; it does not resolve the fundamental need for consent before using customer data for training.

Practice this question →
56
MCQeasy

A nonprofit uses an AI system to allocate resources to communities in need. The system uses historical data which shows that certain neighborhoods have lower service usage. What ethical risk should be considered?

A.The system may violate data minimization principles
B.The system cannot be held accountable for decisions
C.The system lacks explainability
D.The system may perpetuate historical inequities
AnswerD

Using biased historical data can reinforce past discrimination.

Why this answer

The AI system uses historical data that reflects lower service usage in certain neighborhoods. If that historical data is biased due to past inequities (e.g., redlining, underinvestment, or systemic discrimination), the model will learn and amplify those patterns, leading to unfair resource allocation that perpetuates historical disadvantages. This is a classic case of algorithmic bias where the training data encodes societal biases, and the model's predictions reinforce them.

Exam trap

Salesforce often tests the distinction between bias from training data (Option D) versus model explainability (Option C), so candidates mistakenly pick 'lack of explainability' when the real issue is that the model is accurately learning from flawed historical data.

How to eliminate wrong answers

Option A is wrong because data minimization principles (from GDPR and privacy frameworks) concern collecting only necessary personal data, not the fairness of outcomes; the risk here is about bias, not data collection scope. Option B is wrong because AI systems can be held accountable through governance frameworks, audit trails, and human oversight; the statement confuses technical accountability with legal liability. Option C is wrong because while lack of explainability (black-box models) is a concern, the primary ethical risk in this scenario is that the system will replicate historical bias from the training data, not that its decisions are opaque.

Practice this question →
57
MCQmedium

An AI Associate is asked to build a model that predicts employee performance. The dataset includes gender, department, and tenure. Which practice could introduce ethical risk?

A.Evaluating model performance across different groups.
B.Excluding gender from the model features.
C.Documenting model limitations and assumptions.
D.Including gender to improve model accuracy.
AnswerD

Using protected attributes can lead to biased outcomes.

Why this answer

Including gender as a feature in a predictive model for employee performance can introduce bias and lead to unfair or discriminatory outcomes. Even if the model's accuracy improves, using protected attributes like gender may violate ethical guidelines and regulations such as GDPR or anti-discrimination laws, as it could perpetuate historical biases or result in disparate impact.

Exam trap

Salesforce often tests the misconception that including more features always improves model performance, without considering the ethical implications of using protected attributes like gender.

How to eliminate wrong answers

Option A is wrong because evaluating model performance across different groups is a standard fairness practice, such as measuring demographic parity or equal opportunity, and helps identify bias rather than introducing ethical risk. Option B is wrong because excluding gender from the model features is a common bias mitigation technique, often called 'fairness through unawareness,' which reduces the risk of direct discrimination. Option C is wrong because documenting model limitations and assumptions is a responsible AI practice that promotes transparency and accountability, not an ethical risk.

Practice this question →
58
MCQhard

Refer to the exhibit. An AI loan approval policy is defined as a JSON rule set. Which ethical issue is most prominent based on this policy?

A.Use of irrelevant attributes like income and credit score
B.Insufficient accuracy due to simple rules
C.Potential for geographic discrimination due to zip code condition
D.Lack of transparency in decision-making
AnswerC

Zip code can be a proxy for race or socioeconomic status, leading to discrimination.

Why this answer

The use of zip_code as an approval condition can lead to geographic discrimination (redlining). Option A is wrong because the rules are transparent (explicitly shown). Option B is wrong because income and credit score may be relevant, but zip code is problematic.

Option D is wrong because the rules are defined, but accuracy is not directly addressed.

Practice this question →
59
MCQhard

A financial services firm uses Einstein Next Best Action to offer credit products. The model recommends high-interest loans more often to minority groups. The AI Associate must mitigate this. What is the most effective approach?

A.Remove the model and use a rule-based system.
B.Use SHAP values to explain predictions.
C.Apply post-processing fairness adjustments to the recommendations.
D.Add a disclaimer that recommendations may be biased.
AnswerC

This can equalize outcomes without full retraining.

Why this answer

Post-processing fairness adjustments directly modify the model's output to enforce demographic parity or equal opportunity, reducing biased recommendations without retraining the model. This approach is practical when the firm cannot easily change the underlying training data or model architecture, and it allows the AI Associate to intervene at the decision point to ensure fair lending practices.

Exam trap

The trap here is that candidates confuse explainability (SHAP values) with mitigation, thinking that understanding why bias occurs is sufficient to fix it, when in fact only direct adjustments to the model's output can change the biased recommendations.

How to eliminate wrong answers

Option A is wrong because removing the model and using a rule-based system would discard the predictive power of machine learning and likely still encode human biases in the rules, failing to address the root cause of bias. Option B is wrong because SHAP values only explain why a model made a particular prediction; they do not change the model's behavior or mitigate bias in the recommendations. Option D is wrong because adding a disclaimer does not alter the biased outcomes; it merely informs users of potential bias, which does not satisfy regulatory or ethical obligations to ensure fair treatment.

Practice this question →
60
MCQeasy

A Salesforce customer uses Einstein Sentiment Analysis to analyze customer feedback. They find the model is less accurate for non-English languages. What ethical concern does this raise?

A.Bias
B.Accountability
C.Privacy
D.Security
AnswerA

Correct. The model is biased against non-English languages.

Why this answer

The varying accuracy across languages indicates bias in the model, which is a fairness concern.

Practice this question →
61
MCQhard

A company uses Einstein Prediction Builder to predict customer churn. They notice the model is less accurate for a certain segment. What is the best approach to mitigate bias?

A.Increase model complexity
B.Add more features
C.Remove the segment from training
D.Retrain with balanced data
AnswerD

Correct. Balanced data helps the model perform consistently across segments.

Why this answer

Retraining with balanced data directly addresses the root cause of bias: an imbalanced training set where the model underperforms for a specific segment. By ensuring the segment is adequately represented, the model learns more equitable patterns, reducing bias without sacrificing overall accuracy. This aligns with ethical AI practices in Einstein Prediction Builder, where data quality and representation are critical for fair predictions.

Exam trap

Salesforce often tests the misconception that bias is a technical problem solvable by adding complexity or features, when in fact it is a data representation issue requiring balanced training data.

How to eliminate wrong answers

Option A is wrong because increasing model complexity (e.g., adding more layers or interactions) can exacerbate overfitting and may amplify existing biases rather than mitigate them, especially if the biased segment is underrepresented. Option B is wrong because adding more features does not guarantee bias reduction; irrelevant or proxy features can introduce new biases or reinforce existing ones, and the core issue is data imbalance, not feature insufficiency. Option C is wrong because removing the segment from training eliminates the model's ability to predict for that segment entirely, which is a form of exclusion bias and violates ethical AI principles of fairness and inclusivity.

Practice this question →
62
Multi-Selecthard

An AI system used for recruitment has been found to be biased. Which THREE steps should be taken to address this? (Choose three.)

Select 3 answers
A.Deploy the model without changes
B.Audit the training data for bias
C.Retrain the model with a balanced dataset
D.Remove demographic data from the model
E.Monitor outcomes for disparate impact
AnswersB, C, E

Correct. Auditing identifies sources of bias.

Why this answer

Auditing training data, retraining with balanced data, and monitoring outcomes are essential corrective actions.

Practice this question →
63
Multi-Selectmedium

Which two actions are consistent with Salesforce's ethical AI principles when deploying a custom AI model on Salesforce?

Select 2 answers
A.Use only structured data for training.
B.Use the model to make decisions without human review.
C.Optimize for accuracy over all other metrics.
D.Document the model's intended use and limitations.
E.Provide a mechanism for users to challenge model decisions.
AnswersD, E

Transparency is a core ethical principle.

Why this answer

Documenting intended use (transparency) and providing a challenge mechanism (accountability) align with ethical AI principles.

Practice this question →
64
MCQmedium

A developer creates a custom AI model using Salesforce's AI platform. They want to ensure the model is fair. What should they do first?

A.Use a pre-trained model
B.Test on a small sample
C.Collect diverse training data
D.Deploy and monitor
AnswerC

Correct. Diverse data helps prevent systemic bias.

Why this answer

Collecting diverse training data is the foundational step to ensure fairness in an AI model because it helps mitigate bias at the source. Without diverse data representing all relevant groups, the model may learn skewed patterns that lead to discriminatory outcomes, regardless of subsequent testing or monitoring.

Exam trap

Salesforce often tests the misconception that fairness can be achieved through post-hoc actions like testing or monitoring, rather than through proactive data collection, leading candidates to choose 'Test on a small sample' or 'Deploy and monitor' as the first step.

How to eliminate wrong answers

Option A is wrong because using a pre-trained model does not guarantee fairness; the pre-trained model may itself contain biases from its original training data, and the developer still needs to ensure the data used for fine-tuning or adaptation is diverse. Option B is wrong because testing on a small sample is insufficient to detect systemic bias; a small sample may not capture the full range of demographic or behavioral variations, leading to false confidence in fairness. Option D is wrong because deploying and monitoring comes after the model is built; without first ensuring diverse training data, monitoring will only detect issues after harm may have occurred, rather than preventing them at the source.

Practice this question →
65
MCQmedium

A company's Einstein Sentiment model is used to flag negative customer feedback. The model was trained on English reviews only. When deployed globally, it misclassifies positive reviews in Spanish as negative. What is the primary ethical concern?

A.The model is not interpretable.
B.The model has low accuracy for Spanish reviews.
C.The model is unfair to Spanish-speaking customers.
D.The model violates privacy regulations.
AnswerC

Lack of representation leads to unfair treatment.

Why this answer

The primary ethical concern is fairness: the model was trained exclusively on English reviews, so it systematically misclassifies Spanish positive feedback as negative. This creates a disparate impact on Spanish-speaking customers, violating the principle of algorithmic fairness. The issue is not just low accuracy but an unjust bias that disadvantages a specific linguistic group.

Exam trap

Salesforce often tests the distinction between a model's technical flaw (low accuracy) and the ethical principle it violates (fairness), tricking candidates into picking the symptom over the root ethical concern.

How to eliminate wrong answers

Option A is wrong because interpretability (explainability) is a separate concern; the model could be interpretable yet still unfair. Option B is wrong because low accuracy is a symptom, not the primary ethical concern—the core issue is the unfair bias against Spanish speakers, not mere performance degradation. Option D is wrong because the scenario involves no personal data collection or processing that would violate privacy regulations like GDPR or CCPA; the model only analyzes review text for sentiment.

Practice this question →
66
MCQmedium

A company deployed an AI chatbot for customer service. After a week, they receive complaints that the chatbot responds differently based on customer accent. The ethical issue is most likely due to:

A.Lack of personality in the chatbot responses
B.Insufficient computational resources allocated to the chatbot
C.Poor user interface design
D.Bias in the training data leading to discriminatory behavior
AnswerD

Bias in data is a common source of unfair AI behavior.

Why this answer

The chatbot's differing responses based on accent indicate bias in the training data or model. Option D (bias in training data leading to discriminatory behavior) is correct because AI models learn from data, and if the data contains accents correlated with negative outcomes, the model perpetuates that. Option A (insufficient compute resources) is unrelated.

Option B (lack of chatbot personality) is not ethical. Option C (user interface design) is not the cause.

Practice this question →
67
MCQeasy

A government agency uses an AI system to allocate resources for public services such as healthcare and education. The system is designed to optimize overall efficiency based on historical usage data. After deployment, it becomes clear that underserved regions with less historical data receive significantly less funding than well-served regions. The agency's mission is to promote equity. The system's performance metrics show high efficiency, but community leaders protest the unfair distribution. What should the agency do?

A.Abandon the AI system and use a manual, rule-based allocation system.
B.Redesign the system to include fairness constraints that ensure minimum resource levels for underserved regions.
C.Collect more historical data from underserved regions before making adjustments.
D.Continue using the system as is, since it maximizes efficiency.
AnswerB

Fairness constraints balance efficiency with equity, meeting both goals.

Why this answer

Incorporating fairness constraints ensures equitable distribution while still using AI to optimize. Option A ignores the fairness issue. Option C is good but may not be sufficient if the model still biases against underrepresented areas.

Option D reverts to a less efficient system.

Practice this question →
68
MCQmedium

An AI model for predicting employee performance is found to have a higher false positive rate for women than for men. What is the best course of action?

A.Investigate the cause and retrain the model to reduce bias
B.Lower the decision threshold for women to equalize false positive rates
C.Proceed with deployment because the overall accuracy is acceptable
D.Use the model but require manual review for all female candidates
AnswerA

Retraining with fairness constraints mitigates bias.

Why this answer

A higher false positive rate for women indicates the model has learned biased patterns from the training data, likely due to imbalanced or skewed historical data. Investigating the cause—such as examining feature correlations, data distribution, and model architecture—allows for targeted retraining (e.g., reweighting, adversarial debiasing, or fairness constraints) to reduce bias without sacrificing overall performance. This aligns with ethical AI principles and regulatory expectations, ensuring the model is fair across demographic groups.

Exam trap

Salesforce often tests the misconception that adjusting thresholds or adding manual review can fix bias, when in fact these are superficial patches that do not address the root cause in the model's training data or architecture.

How to eliminate wrong answers

Option B is wrong because lowering the decision threshold for women artificially equalizes false positive rates but does not address the underlying bias; it may increase false negatives for women or degrade overall model calibration, leading to inconsistent and unfair outcomes. Option C is wrong because proceeding with deployment despite known bias violates fairness standards and can lead to discriminatory practices, legal liability, and reputational damage, even if overall accuracy is acceptable. Option D is wrong because requiring manual review for all female candidates introduces a separate, potentially biased human-in-the-loop process that is inefficient, costly, and does not fix the model's bias; it also creates a two-tier system that may still result in unfair treatment.

Practice this question →
69
MCQmedium

Refer to the exhibit. A Salesforce developer configures the Einstein Trust Layer as shown. What is the primary purpose of enabling data masking?

A.To improve the accuracy of sentiment analysis.
B.To reduce latency of the AI response.
C.To anonymize personally identifiable information (PII) in the model output.
D.To comply with Salesforce's service-level agreement.
AnswerC

The maskFields specify PII types to be hidden.

Why this answer

Enabling data masking in the Einstein Trust Layer ensures that personally identifiable information (PII) is anonymized before the model output is returned to the user. This protects sensitive data from exposure in AI-generated responses, which is a core requirement for privacy compliance and responsible AI use.

Exam trap

Salesforce often tests the distinction between data masking (which protects output privacy) and data encryption (which protects data in transit or at rest), leading candidates to confuse masking with security controls that affect latency or compliance with SLAs.

How to eliminate wrong answers

Option A is wrong because data masking does not improve sentiment analysis accuracy; it removes or obscures PII, which could actually reduce context for sentiment models if not handled carefully. Option B is wrong because data masking adds processing overhead to scan and redact PII, which may increase latency rather than reduce it. Option D is wrong because while data masking helps meet privacy regulations, it is not specifically tied to Salesforce's service-level agreement (SLA), which covers uptime and performance, not data anonymization.

Practice this question →
70
MCQeasy

Refer to the exhibit. This JSON snippet is from the Einstein Trust Layer configuration. What is the purpose of this configuration?

A.To detect biased predictions based on gender and race
B.To block all predictions involving gender or race
C.To anonymize gender and race data
D.To remove gender and race from the model
AnswerA

Correct. The bias detection feature checks for disparities along these attributes.

Why this answer

The configuration enables bias detection on the specified sensitive attributes (gender and race).

Practice this question →
71
MCQmedium

A healthcare provider uses an AI model to predict patient readmission risk. The model is trained on historical data that underrepresents minority populations. What is the MOST significant ethical risk?

A.The model may overfit to the majority population
B.The model cannot scale to real-time predictions
C.The model may produce biased predictions against minorities
D.The model lacks explainability
AnswerC

Underrepresentation in training data causes algorithmic bias, an ethical risk.

Why this answer

Training on historical data that underrepresents minority populations leads to a model that has insufficient examples to learn patterns for those groups, resulting in biased predictions that systematically disadvantage minorities. This is a direct violation of fairness in AI ethics, as the model's outputs will be less accurate or equitable for underrepresented groups, potentially causing harm in critical healthcare decisions like readmission risk assessment.

Exam trap

Salesforce often tests the distinction between a technical symptom (like overfitting) and the core ethical consequence (like biased predictions), so candidates may incorrectly choose overfitting as the most significant risk instead of recognizing that the ethical harm to minorities is the primary concern.

How to eliminate wrong answers

Option A is wrong because overfitting to the majority population is a symptom of the data imbalance, but the most significant ethical risk is the resulting bias and harm to minority groups, not the overfitting itself. Option B is wrong because the ability to scale to real-time predictions is a performance or deployment concern, not an ethical risk; the model could still be deployed in real-time while producing biased outputs. Option D is wrong because while lack of explainability can be an ethical concern, it is not the most significant risk here; the primary issue is the biased predictions caused by underrepresented data, which can occur even if the model is fully explainable.

Practice this question →
72
MCQhard

Refer to the exhibit. An AI model's accuracy is shown for four demographic groups. Which group should be investigated for potential bias?

A.Group Beta
B.Group Alpha
C.Group Gamma
D.Group Delta
AnswerA

Correct. The low accuracy suggests bias issues.

Why this answer

Group Beta has significantly lower accuracy, indicating possible bias or underperformance.

Practice this question →
73
Multi-Selecteasy

Which TWO practices help ensure accountability in AI systems?

Select 2 answers
A.Blame the AI system for mistakes to protect employees.
B.Assign a human owner for each AI system.
C.Automate all decisions to eliminate human error.
D.Implement logging and auditing of model decisions.
E.Open-source the model code to share responsibility.
AnswersB, D

Human ownership ensures responsibility.

Why this answer

Assigning a human owner for each AI system establishes clear accountability, ensuring that a specific individual is responsible for the system's behavior, decisions, and compliance with ethical guidelines. This practice aligns with the principle of human oversight, which is critical for maintaining trust and addressing failures in AI systems.

Exam trap

Salesforce often tests the distinction between technical transparency (like open-sourcing code) and operational accountability (like assigning a human owner), leading candidates to mistakenly choose open-sourcing as a sufficient accountability practice.

Practice this question →
74
MCQeasy

An organization uses an AI-powered resume screening tool to shortlist candidates for a software engineering role. The tool was trained on historical hiring data from the past five years, during which the company predominantly hired male candidates. After deployment, the tool consistently ranks female candidates lower, even when they have equivalent qualifications. The AI team reports that the overall model accuracy is 92%, and they argue that performance is strong. However, the diversity and inclusion team raises ethical concerns about gender bias. The Salesforce AI Associate is asked to evaluate the situation. What should the associate recommend?

A.Continue using the model because 92% accuracy is acceptable and the bias is not significant.
B.Retrain the model using a balanced dataset that includes equal representation from all genders and implement ongoing fairness monitoring.
C.Replace the current AI tool with a different vendor's tool without further analysis.
D.Manually adjust the scoring algorithm to give preference to female candidates to balance the outcome.
AnswerB

This directly addresses the bias by ensuring the training data is representative and includes measures to monitor fairness.

Why this answer

Retraining with a balanced dataset addresses the root cause of bias, and ongoing monitoring ensures fairness over time. Option A is incorrect because ignoring ethical concerns for accuracy is unacceptable. Option C is incorrect because switching vendors without understanding the bias may not solve the issue.

Option D is incorrect because manually adjusting scores introduces reverse discrimination and is unethical.

Practice this question →
75
MCQhard

Refer to the exhibit. The fairness evaluation shows a disparate impact of 0.85, equal opportunity difference of 0.12, and demographic parity difference of 0.18. Which fairness thresholds are violated?

A.Demographic parity only.
B.Equal opportunity only.
C.Equal opportunity and demographic parity.
D.Disparate impact only.
AnswerC

Both exceed their thresholds.

Why this answer

The fairness evaluation shows violations of both equal opportunity and demographic parity thresholds. The equal opportunity difference of 0.12 exceeds the commonly accepted threshold of 0.1, and the demographic parity difference of 0.18 exceeds the typical threshold of 0.1. Disparate impact of 0.85 is within the acceptable range (typically 0.8 to 1.25), so it is not violated.

Exam trap

Salesforce often tests the misconception that disparate impact is the only fairness metric that matters, but the trap here is that multiple fairness thresholds can be violated simultaneously, and candidates must check each metric against its specific threshold rather than assuming a single violation.

How to eliminate wrong answers

Option A is wrong because demographic parity difference of 0.18 exceeds the 0.1 threshold, but equal opportunity difference of 0.12 also exceeds its 0.1 threshold, so both are violated, not just demographic parity. Option B is wrong because equal opportunity difference of 0.12 exceeds the 0.1 threshold, but demographic parity difference of 0.18 also exceeds its threshold, so both are violated, not just equal opportunity. Option D is wrong because disparate impact of 0.85 falls within the acceptable range of 0.8 to 1.25 (or 80% to 125% rule), so it is not violated.

Practice this question →
Page 1 of 2 · 85 questions totalNext →

Ready to test yourself?

Try a timed practice session using only Ethical Considerations of AI questions.

Start 20-question session