AI Associate Ethical AI and Data Privacy Practice Test 3 — 15 Questions

Question 1

A sales operations manager wants to use Einstein Lead Scoring to prioritize leads. They have historical data showing that leads from a certain postal code have a low conversion rate. However, they suspect the low conversion is due to a past marketing campaign that was poorly targeted, not the demographics of that area. What is the BEST way to ensure the AI model does not unfairly penalize leads from that postal code?

Accepted Answer

Audit the model for disparate impact on that postal code and retrain with updated labels that reflect the true conversion potential. Bias in historical data can lead to unfair predictions. The best approach is to audit the model for bias and retrain with corrected labels or additional features that capture the true drivers of conversion, rather than simply omitting the feature or adjusting scores manually.

Answer

Manually increase the lead scores for all leads from that postal code

Answer

Remove the postal code field from the model training data entirely

Answer

Use the model as-is because Salesforce AI is certified to be fair

Question 2

A Salesforce admin wants to ensure that customer data used by Einstein features is not retained by Salesforce to train foundation models. Which component of the Einstein Trust Layer enforces this commitment?

Accepted Answer

Zero Data Retention. Zero Data Retention is the Einstein Trust Layer policy that ensures customer data is not used to train or improve Salesforce's base AI models. It is a core part of Salesforce's responsible AI framework.

Answer

Audit trail

Answer

Toxicity detection

Answer

PII masking

Question 3

A company deploys an AI-powered email composer for sales reps. The legal team requires that every AI-generated email be reviewed by a human before sending to a customer. Which approach aligns with Salesforce's Trusted AI Principle of Empathy and Human Oversight?

Accepted Answer

Require the sales rep to click 'Approve' before the email is sent. Option D is correct because it directly implements human oversight by requiring the sales rep to click 'Approve' before the email is sent, aligning with Salesforce's Trusted AI Principle of Empathy and Human Oversight. This principle mandates that AI systems should include mechanisms for human review and control, especially in high-stakes communications, ensuring that AI-generated content is vetted for accuracy, tone, and compliance before reaching customers. The approval step creates a clear human-in-the-loop checkpoint, preventing automated sending without human judgment.

Answer

Allow the AI to send emails automatically but log all sent emails for audit

Answer

Use a rule to block specific words in AI-generated emails

Answer

Send a daily report of all AI-generated emails to the legal team after they are sent

Question 4

A data scientist is building a churn prediction model using Einstein Discovery. They want to ensure the model does not rely on sensitive attributes like race or gender, even if those are correlated with other features. Which technique is MOST aligned with Salesforce's data minimisation principle?

Accepted Answer

Remove sensitive attributes from the training data and avoid using proxies that strongly correlate with them. Data minimisation means using only the data necessary for the task. Excluding sensitive attributes from the feature set is the most direct way to prevent them from being used, even if they are correlated with other features. Correlation does not imply causation, and if those features are not essential, they should be removed.

Answer

Include all features and rely on the AI to ignore biased ones

Answer

Apply a fairness constraint after training to adjust predictions

Answer

Use differential privacy to add noise to the training data

Question 5

A financial services company uses Einstein Bots to answer customer inquiries. A customer asks the bot to explain why their loan application was rejected. The bot provides a response based on AI predictions. Which Salesforce Trusted AI Principle is MOST directly addressed by the bot's ability to explain the decision?

Accepted Answer

Transparency. Transparency in AI means that decisions should be explainable and understandable to users. Providing an explanation for a loan rejection directly supports the principle of Transparency, which requires AI systems to be open about their reasoning.

Answer

Honesty

Answer

Accuracy

Answer

Safety

Question 6

A company wants to use Einstein Sentiment Analysis to classify customer feedback. What is the FIRST step they should take to ensure ethical use of customer data?

Accepted Answer

Obtain proper consent from customers for using their data in AI. Ethical AI begins with data governance. Obtaining proper consent from customers to use their feedback for AI analysis is a fundamental requirement under privacy regulations like GDPR and CCPA. It also aligns with the Trusted AI principle of Empathy, respecting customer autonomy.

Answer

Set up an audit trail for all AI actions

Answer

Choose a pre-trained model to avoid using customer data

Answer

Anonymize all customer feedback before analysis

Question 7

A sales manager notices that Einstein Lead Scoring assigns lower scores to leads from a specific region. After investigation, they find that the historical conversion data for that region is sparse and unrepresentative. What should the manager do to improve the model's fairness?

Accepted Answer

Collect more data from that region to provide a representative sample for retraining. Option A is correct because the core issue is data sparsity and unrepresentative historical conversion data for that region. By collecting more data from that region, the model can be retrained on a balanced dataset, which directly addresses the root cause of the biased lead scoring. This aligns with the principle of fairness in AI, where models must be trained on representative data to avoid systematic discrimination.

Answer

Adjust the lead scoring formula manually to increase scores for that region

Answer

Remove the region field from the model

Answer

Accept the scores as-is because the model is certified

Question 8

A healthcare organization uses Einstein Next Best Action to recommend treatments to patients. They must comply with GDPR's right to explanation for automated decisions. Which combination of Einstein Trust Layer features is MOST essential to meet this requirement?

Accepted Answer

Score factors and grounding. The right to explanation requires that individuals can obtain meaningful information about the logic involved in automated decisions. Score factors provide the most influential fields for a prediction, and grounding connects the AI to relevant CRM data, enabling a transparent explanation. Toxicity detection is not relevant. Audit trail logs actions but does not provide explanation to the patient.

Answer

Audit trail and PII masking

Answer

Toxicity detection and audit trail

Answer

Zero Data Retention and PII masking

Question 9

A company is deploying Einstein Article Recommendations on its customer portal. They want to ensure customers know that recommendations are AI-generated. Which action aligns with the Salesforce Trusted AI Principle of Honesty?

Accepted Answer

Include a note stating 'Recommended by Einstein AI' on each recommendation. Option C is correct because the Salesforce Trusted AI Principle of Honesty requires transparency about AI-generated content. By explicitly stating 'Recommended by Einstein AI' on each recommendation, the company clearly discloses that the recommendations are AI-generated, building trust with customers. This aligns with the principle's focus on avoiding deception and ensuring users understand the nature of the content they are seeing.

Answer

Provide a chatbot that can explain the recommendations

Answer

Use a generic label like 'Recommended for you'

Answer

Do not mention AI; just show the recommendations

Question 10

A company wants to use Einstein Prediction Builder to predict which customers are likely to churn. They have a dataset that includes customers' names, email addresses, and detailed purchase history. According to the data minimisation principle, which fields should be included in the model?

Accepted Answer

Only the purchase history. Data minimisation means using only the data necessary for the prediction. Purchase history is relevant to churn prediction, while names and email addresses are personally identifiable information (PII) that are not needed for the model and should be excluded to minimize privacy risks.

Answer

Only the email addresses, since they are unique identifiers

Answer

All three fields because more data improves accuracy

Answer

Names and purchase history, but not email addresses

Question 11

A customer service manager wants to use Einstein Bots to handle common inquiries. They are concerned about the bot generating offensive responses. Which Einstein Trust Layer feature should they enable to minimize this risk?

Accepted Answer

Toxicity detection. Toxicity detection is a feature of the Einstein Trust Layer that identifies and filters harmful or offensive language before it is sent to customers. This directly addresses the concern about offensive responses, aligning with the Safety principle.

Answer

Zero Data Retention

Answer

Grounding

Answer

PII masking

Question 12

A company uses Einstein Discovery to build a model that predicts customer lifetime value. They want to ensure the model's decisions are auditable and that they can track the business impact of AI recommendations over time. What should they implement?

Accepted Answer

Audit trail. An audit trail logs when AI recommendations are made and whether they were followed, and can be used to track outcomes over time. This enables accountability and measurement of AI impact, supporting transparency and governance.

Answer

Score factors

Answer

Grounding

Answer

Toxicity detection

Question 13

A company is implementing Einstein Next Best Action for their customer service agents. They want to ensure that AI recommendations are provided as suggestions but that agents retain the final say. Which TWO practices support this goal of human oversight?

Accepted Answer

Configure the system to require agent confirmation before executing any AI-recommended action. Human oversight can be achieved by requiring agents to confirm before an action is taken, and by clearly presenting AI recommendations as suggestions. Disabling the AI or automating actions removes human involvement. Training agents is helpful but does not directly ensure oversight in the workflow.

Answer

Disable all AI recommendations and rely solely on agent judgment

Answer

Train all agents on how AI models work

Answer

Automate the top recommendations to save agent time

Question 14

A company must comply with GDPR when using Einstein Features to process customer data for AI predictions. Which THREE actions are required under GDPR that relate directly to AI-driven decisions?

Accepted Answer

Ensure that only data necessary for the prediction is used (data minimisation). GDPR requires a lawful basis for processing (e.g., consent), the right to explanation for automated decisions, and data minimization. Data retention policy is important but not exclusive to AI decisions; and bias auditing is a best practice but not a GDPR requirement.

Answer

Implement a data retention policy that deletes customer data after 30 days

Answer

Audit AI models for bias on a quarterly basis

Question 15

A sales team uses Einstein Lead Scoring and notices that the model gives disproportionately low scores to leads from a certain demographic group. The team suspects historical bias in the training data. Which THREE steps should they take to address this bias?

Accepted Answer

Investigate the historical data to identify and correct labeling errors or sampling bias. To address bias, the team should audit the model for disparate impact, investigate and correct the historical data, and consider adding features that reduce reliance on biased proxies. Removing the model entirely is unnecessary, and manually adjusting scores is not a systemic fix.

Answer

Manually increase scores for the affected group by a fixed percentage

Answer

Remove the model and rely on manual lead scoring