AI Associate Practice Exam 57 — 20 Questions

Question 1

A retail company uses Salesforce Einstein Vision to analyze customer images for product recommendations. The AI team notices that the model performs poorly on images of customers with darker skin tones, leading to fewer recommendations for that demographic. The team has access to a dataset of diverse skin tones but the company's data privacy policy prohibits using demographic data in training. What should the team do?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Document the bias, escalate to the ethics board, and seek guidance on using diverse data while maintaining privacy.. Option C is correct because when a model shows demographic performance disparity and privacy policy blocks the obvious fix, the proper response is to document the bias, escalate to the ethics board, and obtain governance guidance on how to use diverse data compliantly. This preserves accountability and ensures any mitigation respects the data privacy policy rather than unilaterally overriding it. Option A is wrong because retraining on the diverse dataset without considering privacy regulations violates the stated policy. Option B is wrong because ignoring the disparity leaves the discriminatory performance unaddressed. Option D is wrong because simply removing skin tone labels does not resolve the privacy issue or guarantee the bias is mitigated, since the underlying demographic data would still be used without approval.

Answer

Retrain the model on the diverse dataset without considering privacy regulations.

Answer

Ignore the performance disparity as it is a result of natural data distribution.

Answer

Use the diverse dataset but remove skin tone labels to avoid privacy issues.

Question 2

A Salesforce developer is building an Einstein Bots-powered customer service chatbot. The bot will handle billing inquiries and may access customers' account details. Which design principle best ensures ethical handling of customer data?

Accepted Answer

Implement data minimization so the bot only retrieves information necessary to resolve the specific inquiry.. Data minimization ensures the chatbot only accesses information required to resolve the inquiry, reducing privacy risks and complying with ethical and legal standards. This principle is central to responsible AI, as it prevents unnecessary exposure of customer data. By limiting data retrieval, the organization demonstrates respect for customer privacy and builds trust.

Answer

Store all conversation logs indefinitely to improve future bot responses.

Answer

Use the Einstein Trust Layer to mask all customer data before the bot processes it.

Answer

Enable the bot to access all customer records to provide comprehensive answers quickly.

Question 3

A small business uses a pre-built Salesforce AI model to predict inventory needs. The model recommends ordering extra stock based on seasonal trends. One month, the model fails to predict a sudden demand spike, resulting in stockouts and lost sales. The business owner is frustrated and considers disabling the AI. The owner wants to know if this is an ethical issue and what to do next. As an AI ethics advisor, what is the best response?

Accepted Answer

Explain that this is a model performance issue, not an ethics violation, and suggest reviewing the model's accuracy and retraining with recent data. Option C is correct because a missed demand spike from a pre-built Salesforce AI model is a model performance and accuracy problem, not an ethics violation; the appropriate next step is to evaluate the model's accuracy metrics and retrain or update it with recent, more representative data. Ethical issues would involve concerns such as bias, privacy, transparency, or misuse, none of which are described here. Option A is wrong because disabling the AI immediately is an overreaction that ignores remediation and business value. Option B is wrong because no AI model can be guaranteed 100% accurate. Option D is wrong because not all AI failures are ethical issues; many are simply performance limitations.

Answer

Agree that the AI is unreliable and recommend disabling it immediately

Answer

Recommend using a different AI model that is guaranteed to be 100% accurate

Answer

State that all AI failures are ethical issues and the company should stop using AI

Question 4

A Salesforce admin is setting up a dataset for an Einstein Prediction Builder model to predict whether a lead will convert. The dataset includes a field 'Industry' that has many missing values. What is the recommended approach to handle missing values in this categorical field?

Accepted Answer

Replace missing values with the most frequent industry in the dataset.. For categorical fields with missing values, imputing with the most frequent category (mode) is a recommended practice. It retains data and maintains distribution. Deleting records or leaving blanks can cause bias or errors, while arbitrary numeric conversion introduces false relationships. This approach ensures the model can use the 'Industry' field effectively.

Answer

Delete all records with missing 'Industry' values to ensure data completeness.

Answer

Convert the 'Industry' field to a numeric field by assigning arbitrary numbers to each industry.

Answer

Leave the missing values as blank because Einstein Prediction Builder automatically handles nulls.

Question 5

A Salesforce admin is preparing a dataset for an Einstein Discovery model that predicts customer churn. The dataset includes a field named 'CustomerID' that uniquely identifies each customer, plus several numeric and categorical fields. The admin wants to ensure the model learns meaningful patterns without memorizing individual customers. What should the admin do with the 'CustomerID' field?

Accepted Answer

Exclude the 'CustomerID' field from the model because it is an identifier and not a predictive feature.. Unique identifiers like CustomerID do not contain predictive information and can cause a model to memorize individual records, which harms generalization. Excluding such fields ensures the model learns from actual behavioral and demographic attributes that correlate with churn. This practice improves model robustness and avoids overfitting.

Answer

Use 'CustomerID' as the outcome variable to predict which customers will churn.

Answer

Convert 'CustomerID' to a numeric value and include it as a feature to help the model differentiate customers.

Answer

Keep 'CustomerID' in the dataset but exclude it during model training only.

Question 6

After deploying an AI model in Salesforce, the data scientist notices high accuracy on the training set but poor accuracy on new incoming data. What is this phenomenon called?

Accepted Answer

Overfitting. Overfitting, which occurs when a model learns the training data too well, including its noise and idiosyncrasies, so it performs excellently on the training set but generalizes poorly to new, unseen data—exactly the high-training-accuracy, low-new-data-accuracy pattern described. Overfitting typically arises from an overly complex model, too many features relative to the number of training examples, or insufficient regularization. Data leakage (A) would inflate training performance via information from outside the training set, but it does not by itself describe the generalization gap as the named phenomenon. Underfitting (C) is the opposite problem, producing poor accuracy on both training and new data. Concept drift (D) refers to the statistical properties of the target changing over time, not to a model that fits training data too closely.

Answer

Data leakage

Answer

Underfitting

Answer

Concept drift

Question 7

A financial institution uses Salesforce Einstein to generate personalized investment advice. The AI model occasionally recommends high-risk products to elderly clients with conservative risk profiles. Which ethical principle is most directly violated?

Accepted Answer

Beneficence, because the AI is not acting in the best interest of the client.. Beneficence is the ethical principle of acting in the best interest of others. The AI's recommendations of high-risk products to conservative elderly clients directly harm their financial well-being, violating this principle. Ethical AI must prioritize user welfare and tailor advice to individual circumstances, ensuring that recommendations are suitable and beneficial.

Answer

Transparency, because the AI does not explain why it recommends high-risk products.

Answer

Privacy, because the AI accesses elderly clients' financial data without consent.

Answer

Justice, because the AI treats elderly clients differently from younger clients.

Question 8

A retail company uses AI to personalize marketing emails. A customer complains that their data was used without explicit permission. What ethical principle was most likely violated?

Accepted Answer

Consent and privacy. Consent and privacy, because the scenario describes a customer whose data was used for AI-driven email personalization without explicit permission, which directly violates the ethical requirement to obtain informed consent and protect personal data. Consent and privacy principles require organizations to be transparent about data collection and to give individuals control over how their information is used, especially for marketing purposes. Transparency (A) relates to disclosing how data is used, but the core violation here is the absence of permission, not merely a lack of disclosure. Fairness (B) concerns avoiding bias and equitable treatment, which is not the issue described, and accountability (D) concerns taking responsibility for outcomes, which is secondary to the missing consent.

Answer

Transparency

Answer

Fairness

Answer

Accountability

Question 9

A company wants to build a sentiment analysis model using customer feedback. What is the best practice for labeling the training data?

Accepted Answer

Employ a diverse set of human labelers with clear guidelines. Option C is correct because employing a diverse set of human labelers with clear guidelines reduces individual bias and improves label consistency, which is a best practice for supervised sentiment analysis training data. Clear annotation guidelines ensure subjective sentiment judgments are applied uniformly, and multiple labelers allow measuring inter-annotator agreement to validate quality. Option A is wrong because sentiment analysis typically requires labeled examples for supervised training, and ignoring labels does not meet the stated goal. Option B is weaker than C because a single expert can introduce personal bias and does not provide agreement checks, while Option D is unreliable since keyword matching misses context, sarcasm, and nuanced sentiment.

Answer

Ignore labeling and use unsupervised learning

Answer

Have a single domain expert label all data

Answer

Use automated keyword matching to assign sentiment

Question 10

Refer to the exhibit. The model is deployed and monitoring triggers an alert for a fairness violation. What does this indicate?

Accepted Answer

The demographic parity difference exceeded 0.1.. The alert indicates that the demographic parity difference exceeded 0.1, which is the configured fairness threshold that triggers a fairness violation in Amazon SageMaker Clarify model monitoring. Demographic parity difference measures the absolute difference in positive prediction rates between groups defined by a protected attribute, and a value above 0.1 signals meaningful disparity. The other options do not fit: removing protected attributes (B) is a data-preprocessing action, not a fairness metric alert; retraining the model (C) is an operational event unrelated to the violation; and an accuracy drop below threshold (D) would trigger a model quality or drift alert, not a fairness violation.

Answer

The protected attributes were removed.

Answer

The model has been retrained.

Answer

The model's accuracy has dropped below threshold.

Question 11

Which TWO capabilities are available in Einstein GPT for Sales?

Accepted Answer

Summarizing call transcripts. Option D is correct because Einstein GPT for Sales includes the Sales Summaries / call summary capability, which uses generative AI to summarize call transcripts and surface key points and next steps for sellers. Option E is correct because Einstein GPT for Sales provides Sales Email, which generates personalized email drafts grounded in CRM data such as the recipient's account, opportunity, and prior activity. Option A is not a listed Einstein GPT for Sales capability; sentiment analysis of emails is not one of its generative features. Option B is not correct because automated lead enrichment is a separate data-enrichment function, not an Einstein GPT for Sales capability. Option C is not correct because generating Apex code is a developer-focused capability (e.g., Einstein for Developers), not part of Einstein GPT for Sales.

Answer

Sentiment analysis of customer emails

Answer

Automated lead enrichment

Answer

Generating Apex code

Question 12

A healthcare company uses an AI model built on Salesforce to predict patient readmission risk. The model is trained on historical data that underrepresents certain ethnic groups. During testing, the model shows significantly higher false negative rates for those groups, meaning it fails to flag high-risk patients. The ethical concern is most directly related to which AI principle?

Accepted Answer

Fairness. Fairness, because the model's significantly higher false negative rates for underrepresented ethnic groups mean it delivers unequal outcomes across demographic groups, which is the definition of bias and a fairness violation in AI ethics. Fairness specifically addresses disparate performance or treatment of protected groups, and here the harm is that high-risk patients in those groups are not flagged. Privacy (A) concerns data protection and consent, not unequal model errors. Accountability (B) concerns who is responsible for AI outcomes, and Transparency (C) concerns explainability and disclosure, neither of which directly describes the biased error-rate disparity.

Answer

Privacy

Answer

Accountability

Answer

Transparency

Question 13

A financial services firm deploys Einstein Prediction Builder to predict loan default risk. The model uses sensitive attributes like zip code and age. During testing, the model shows a disparate impact on minority neighborhoods. The compliance team requires explanation of individual predictions for regulatory audits. The data science team wants to use a complex deep learning model that is not interpretable. Which approach best balances performance and ethical responsibility?

Accepted Answer

Use a simpler, interpretable model (e.g., logistic regression) that may have slightly lower accuracy but ensures transparency and reduces bias.. In a regulated financial-services scenario, a simpler interpretable model such as logistic regression provides inherently transparent, auditable individual predictions and makes it easier to detect, document, and mitigate disparate impact on protected groups like minority neighborhoods, even at some cost in accuracy. The compliance requirement for explaining individual predictions is best met by a model whose decision logic is directly inspectable rather than one requiring post-hoc approximations. Options A and B retain the uninterpretable deep learning model, so individual explanations remain approximate and bias risk persists, and limiting exposure does not satisfy the audit requirement. Option C is unethical and likely illegal, since adjusting thresholds per group to conceal disparate impact is discriminatory manipulation rather than a valid fairness remedy.

Answer

Use the complex model but provide post-hoc explanations like SHAP values to satisfy compliance.

Answer

Use the complex model but only for a subset of customers to limit exposure.

Answer

Use the complex model and hide the disparate impact by adjusting thresholds per group.

Question 14

A Salesforce administrator is configuring a Data Cloud data stream to ingest customer profiles from an external CRM. The source sends a 'LastModifiedDate' field in the format 'MM/DD/YYYY', but Data Cloud expects ISO 8601 ('YYYY-MM-DD'). Which Data Cloud feature should be used to transform the date format during ingestion?

Accepted Answer

Data Cloud Data Transform. Data Cloud Data Transform is specifically designed to modify data during ingestion, including date format conversions. Using it ensures the incoming 'MM/DD/YYYY' values are converted to ISO 8601 before being stored in Data Cloud, which is critical for accurate AI model training and consistency.

Answer

Salesforce Flow

Answer

CRM Analytics Data Prep Recipe

Answer

Einstein Prediction Builder

Question 15

A large enterprise uses Data Cloud to power an Einstein model for lead scoring. The model's feature pipeline includes dozens of fields from multiple data streams. Performance has degraded, and the team suspects slow feature retrieval. What is the most efficient way to speed up feature computation in Data Cloud?

Accepted Answer

Use Calculated Insights to pre-compute and cache common features. The correct option is C: Use Calculated Insights to pre-compute and cache common features. Calculated Insights in Data Cloud are designed to compute and persist aggregated or derived metrics (features) ahead of time, so the Einstein model's feature pipeline can retrieve them quickly instead of recalculating them on every scoring request, directly addressing the slow feature retrieval. Option A (increasing data stream parallelism) affects ingestion throughput, not feature computation latency. Option B (external caching in the application layer) adds complexity and does not leverage Data Cloud's native feature computation. Option D (moving all data to a single data lake object) does not optimize feature computation and may even worsen performance by consolidating unrelated streams.

Answer

Increase the parallelism of the data streams

Answer

Implement external caching in the application layer

Answer

Move all data to a single data lake object

Question 16

A Salesforce admin is configuring the Einstein Trust Layer for a generative AI feature that summarizes customer cases. Which two actions should the admin take to uphold ethical AI principles? (Choose two.)

Accepted Answer

Configure audit trail logging to record all prompts and responses for compliance review.. Enabling data masking and audit trail logging are key ethical practices for the Einstein Trust Layer. Data masking protects PII from being sent to external LLMs, while audit logging ensures accountability and transparency. Together, they uphold privacy and compliance, aligning with Salesforce's responsible AI principles.

Answer

Disable all safety guardrails to allow the AI to generate more creative responses.

Answer

Allow the AI to access all customer records to provide more accurate summaries.

Answer

Store all generated summaries in a public repository for transparency.

Question 17

A healthcare company uses AI to predict patient readmission rates. What is a critical ethical requirement?

Accepted Answer

Explanation of predictions to doctors. Explanation of predictions to doctors, because in a healthcare setting clinicians must be able to understand and justify AI-driven readmission risk scores before acting on them, which supports informed consent, accountability, and patient safety. Explainability is a core ethical requirement for high-stakes medical AI under frameworks such as the EU AI Act's high-risk classification and HIPAA-adjacent clinical governance. Low latency (B) is a performance concern, not an ethical mandate, and high precision (C) is a useful metric but does not by itself ensure ethical or transparent use. Using external data sources (D) is neither required nor necessarily ethical, since it can introduce privacy and consent risks rather than resolve them.

Answer

Low latency

Answer

High precision

Answer

Use of external data sources

Question 18

A marketing agency needs to ingest real-time social media mentions for a sentiment analysis AI model. Which Data Cloud object type should they use to set up the ingestion?

Accepted Answer

Data Stream Object with Event type. The correct option is C, a Data Stream Object with Event type, because real-time social media mentions are time-stamped, event-driven data that must be ingested continuously into Data Cloud, and the Event stream type is designed exactly for high-volume, real-time ingestion of streaming data. Data Stream Objects are the ingestion mechanism in Data Cloud, and the Event type supports streaming sources rather than batch loads. Option A, Data Lake Object, is a storage/mapping target for ingested data, not the ingestion setup itself. Option B, Calculated Insight, is used to compute metrics and aggregations for activation, not to ingest raw data. Option D, Data Transform, is used to transform already-ingested data (e.g., via SQL-based recipes), not to establish the real-time ingestion stream.

Answer

Data Lake Object

Answer

Calculated Insight

Answer

Data Transform

Question 19

A social media platform's AI recommends content that inadvertently amplifies misinformation. An ethical review board is considering changes. Which of the following actions best addresses the unintended harm?

Accepted Answer

Conduct an ethical impact assessment and adjust algorithms accordingly. Conducting an ethical impact assessment directly identifies how the recommendation algorithm amplifies misinformation and then adjusts the algorithm to mitigate that specific harm, which is the appropriate governance response for unintended ethical side effects. Increasing training data (A) does not address the underlying objective or feedback loop that promotes misinformation and may even worsen it. Removing all AI recommendation engines (B) is a disproportionate, overbroad response that discards legitimate functionality rather than fixing the harm. Reducing overall user engagement metrics (D) is a vague business change that does not systematically detect or correct the algorithmic cause of misinformation amplification.

Answer

Increase the amount of training data

Answer

Remove all AI recommendation engines

Answer

Reduce overall user engagement metrics

Question 20

A data analyst is preparing a dataset for an Einstein Prediction Builder model that predicts whether a customer will make a repeat purchase. The dataset contains a field 'LastPurchaseDate' with some missing values. The analyst wants to handle these missing values appropriately. What is the recommended approach?

Accepted Answer

Leave missing values as null and let Einstein Prediction Builder handle them automatically.. Einstein Prediction Builder automatically handles missing values, so leaving them as null is appropriate. Imputing with arbitrary values or removing records can introduce bias or reduce data volume. Allowing the platform to manage missing data ensures the model can use all available information and learn from patterns of missingness.

Answer

Impute missing values with a default date such as January 1, 1900, to indicate missingness.

Answer

Remove records with missing 'LastPurchaseDate' from the dataset before training.

Answer

Replace missing values with the average purchase date across all customers.