Courseiva

Salesforce AI Associate AI Associate (AI Associate) — Questions 376–450

506 questions total · 7pages · All types, answers revealed

Take a mock examExam hub
Page 5

Page 6 of 7

Page 7
376
MCQeasy

A company plans to use Einstein Discovery to analyze sales data. Which data preparation step is essential for time-series forecasting?

A.Remove all outliers in sales amounts
B.Ensure date fields are properly formatted and contain sufficient historical range
C.Remove duplicate records
D.Scale all numeric fields to a 0-1 range
AnswerB

Einstein Discovery relies on date fields for trend detection.

Why this answer

For time-series forecasting in Einstein Discovery, the date field must be properly formatted (e.g., as a date or datetime data type) and contain a sufficient historical range to identify patterns like seasonality and trends. Without adequate historical data, the model cannot learn temporal dependencies, making this step essential.

Exam trap

Salesforce often tests the misconception that data normalization (scaling) is always required for AI models, but for tree-based algorithms like those in Einstein Discovery, scaling is irrelevant, and the trap is that candidates pick Option D thinking it is a universal preprocessing step.

How to eliminate wrong answers

Option A is wrong because removing all outliers in sales amounts can discard legitimate extreme values that represent real-world events (e.g., holiday spikes, promotions), which are critical for accurate time-series forecasting; Einstein Discovery handles outliers through model tuning rather than blanket removal. Option C is wrong because while removing duplicate records is a general data cleaning best practice, it is not specifically essential for time-series forecasting; duplicates in date-indexed data are typically handled by aggregation or deduplication, but this step is not a prerequisite for the forecasting algorithm. Option D is wrong because scaling numeric fields to a 0-1 range is unnecessary for time-series forecasting in Einstein Discovery, as tree-based models (like Gradient Boosted Trees) used internally are invariant to monotonic transformations and do not require normalization.

Full explanation →
377
MCQmedium

A news aggregator app uses an AI algorithm to personalize the news feed for each user. The algorithm selects articles based on past clicks and reading time. Recently, a study reveals that the algorithm disproportionately shows sensational and polarizing news to users from certain political orientations, while showing more neutral content to others. The company's user engagement metrics have increased, but journalists express concern about reinforcing echo chambers and misinformation. The company wants to uphold ethical standards while keeping users engaged. What should they do?

A.Modify the algorithm to include diversity and reliability scores for news sources, promoting a balanced feed.
B.Allow users to manually select the types of news they want to see.
C.Show the same generic news feed to all users.
D.Continue with the current algorithm since it increases engagement.
AnswerA

This encourages exposure to different viewpoints and reduces the spread of unreliable content.

Why this answer

Option B is correct because incorporating diversity and reliability metrics into the algorithm balances engagement with ethical responsibilities. Option A ignores ethics. Option C eliminates personalization, hurting engagement.

Option D shifts responsibility to users, which may not be effective.

Full explanation →
378
MCQhard

A retail company uses Einstein to personalize product recommendations. The AI model is trained on customer purchase data that includes sensitive attributes like race and gender. The company wants to ensure ethical use. Which action would best address fairness concerns?

A.Remove race and gender fields from the training dataset
B.Obtain explicit consent from customers for data use
C.Add more demographic data to improve model accuracy
D.Randomize recommendations to ensure equal treatment
AnswerA

Removing protected attributes helps prevent direct discrimination.

Why this answer

The correct answer is A because removing sensitive attributes from training data mitigates direct discrimination. Option B is wrong because adding more data might not remove bias. Option C is wrong because randomizing recommendations reduces relevance and does not address bias.

Option D is wrong because obtaining additional consent does not fix bias in the model.

Full explanation →
379
MCQmedium

A marketing team wants to predict which segment of customers will likely purchase a new product. Which Einstein feature is most appropriate?

A.Custom Reports
B.List Views
C.Campaigns
D.Einstein Segment Prediction
AnswerD

Predicts segment behavior using ML.

Why this answer

Einstein Segment Prediction uses predictive modeling and machine learning to analyze historical customer data and identify which segments are most likely to purchase a new product. It is specifically designed for predictive segmentation based on behavioral and demographic patterns, making it the most appropriate choice for this use case.

Exam trap

Salesforce often tests the distinction between descriptive analytics (reports, list views) and predictive analytics (Einstein features), leading candidates to choose a familiar but incorrect option like Custom Reports or List Views instead of the AI-powered prediction tool.

How to eliminate wrong answers

Option A is wrong because Custom Reports are used for creating ad-hoc data views and analyzing past performance, not for generating predictive insights about future purchase likelihood. Option B is wrong because List Views are static filters that display records based on predefined criteria, lacking any machine learning or predictive capability. Option C is wrong because Campaigns are used to manage marketing outreach and track engagement, not to predict which customer segments will purchase a product.

Full explanation →
380
MCQeasy

A customer service department uses an AI chatbot to handle common inquiries. Recently, customers have reported that the chatbot sometimes responds with offensive or inappropriate language. The company wants to uphold ethical standards. Which approach is the best practice?

A.Implement a filter to automatically block any offensive words in the chatbot's responses.
B.Limit the chatbot to only respond from a fixed set of predefined answers.
C.Implement human review of all chatbot responses that are flagged as potentially offensive.
D.Use a pre-trained language model from a trusted vendor to guarantee ethical behavior.
AnswerC

Human review ensures nuanced handling of sensitive content and allows model improvement based on feedback.

Why this answer

Option B is correct because human review for flagged responses provides a safety net and continuous improvement loop. Option A may block offensive words but can lead to false positives. Option C assumes pre-trained models are unbiased, which is not guaranteed.

Option D reduces functionality and customer experience.

Full explanation →
381
Multi-Selecthard

Which THREE of the following are best practices for training an Einstein Bot?

Select 3 answers
A.Train with only one intent to avoid confusion
B.Test the bot with sample conversations before deployment
C.Use exact match phrases only
D.Use a large and diverse set of training phrases for each intent
E.Include negative examples to improve accuracy
AnswersB, D, E

Testing ensures the bot performs as expected in real scenarios.

Why this answer

Option B is correct because testing an Einstein Bot with sample conversations before deployment allows you to validate the bot's intent recognition, dialog flow, and response accuracy in a controlled environment. This practice helps identify and fix issues with phrase matching, slot filling, and escalation paths, ensuring the bot performs reliably in production.

Exam trap

Salesforce often tests the misconception that more training data is always better, but the trap here is that candidates may overlook the importance of diversity and negative examples, thinking that exact matches or a single intent simplify training, when in fact they cripple the bot's NLP accuracy.

Full explanation →
382
MCQhard

A financial services company is deploying Einstein Prediction Builder to predict customer churn. The data includes both numerical and categorical fields. Which step is essential to ensure the model is not biased against protected attributes like race or gender?

A.Include race and gender as predictors to allow the model to adjust for them.
B.Rely on the model's built-in fairness constraints.
C.Use a deep learning algorithm to automatically handle bias correction.
D.Exclude any protected attributes from the training data and ensure the model does not use correlated proxies.
AnswerD

This is the standard approach to mitigate bias.

Why this answer

Excluding protected attributes like race or gender from the training data and ensuring the model does not use correlated proxies is essential to prevent bias in Einstein Prediction Builder. This approach directly removes the risk of the model learning discriminatory patterns based on these attributes, as the platform relies on the data provided and does not automatically enforce fairness constraints. Including such attributes or relying on built-in fairness would not guarantee unbiased predictions because the model could still infer protected characteristics from correlated features.

Exam trap

Salesforce often tests the misconception that including protected attributes allows the model to 'adjust' for bias, when in reality it introduces direct bias, and that built-in fairness constraints or advanced algorithms can automatically fix bias without explicit data preparation.

How to eliminate wrong answers

Option A is wrong because including race and gender as predictors would allow the model to directly learn and potentially amplify biases, leading to discriminatory outcomes rather than adjusting for them. Option B is wrong because Einstein Prediction Builder does not have built-in fairness constraints that automatically correct for bias; it requires careful data preparation and feature selection by the user. Option C is wrong because deep learning algorithms do not inherently handle bias correction; they can actually exacerbate biases present in the data if not explicitly mitigated through techniques like adversarial debiasing or reweighting.

Full explanation →
383
MCQmedium

During the data preparation phase for an AI model, a data engineer discovers that the 'AnnualRevenue' field contains some negative values. What is the best course of action?

A.Delete all records with negative revenue
B.Replace negative values with the mean of positive values
C.Keep negative values as they might represent returns or refunds
D.Investigate the data source to correct the negative values
AnswerD

Correcting at source ensures data integrity.

Why this answer

Option D is correct because negative revenue values typically indicate data entry errors, system bugs, or incorrect data transformations. The best practice in data preparation is to investigate the source system to understand why negative values were generated and correct them at the origin, ensuring data integrity before any imputation or deletion. Simply deleting or imputing without root-cause analysis can introduce bias or mask underlying data quality issues.

Exam trap

Salesforce often tests the misconception that imputation (e.g., mean replacement) is a safe default for handling invalid data, when in fact the correct first step is always to trace and fix the root cause at the data source.

How to eliminate wrong answers

Option A is wrong because deleting records with negative revenue can introduce selection bias and reduce the dataset size, potentially discarding valid data if negative values represent legitimate business events like refunds. Option B is wrong because replacing negative values with the mean of positive values artificially inflates the central tendency and distorts the distribution, which can degrade model performance, especially for regression tasks. Option C is wrong because keeping negative values as-is without investigation assumes they are valid, but in most financial datasets, revenue is non-negative by definition, and unverified negative values will mislead the model during training.

Full explanation →
384
MCQhard

What is the primary purpose of this policy?

A.Data governance for AI
B.Data integration
C.Data transformation
D.Data backup
AnswerA

Policy controls access, masking, and retention.

Why this answer

Option C is correct because the policy defines data access rules, field allowances, masking, and retention, which are all components of data governance for AI. Option A is wrong because data integration focuses on combining data, not controlling access. Option B is wrong because data backup is about recovery.

Option D is wrong because data transformation changes data format.

Full explanation →
385
MCQhard

A data architect notices that a Data Stream from an external ERP system is failing intermittently with schema mismatch errors. The ERP team says the schema changes occasionally. What is the most effective long-term solution?

A.Implement Data Stream schema validation and flexible mapping
B.Use a different primary key for the data model
C.Increase the number of retries for the Data Stream
D.Ask the ERP team to manually notify of changes
AnswerA

Automates handling of schema changes.

Why this answer

Option C is correct because adding schema validation and mapping checks can detect and handle changes gracefully. Option A is wrong because increasing retries doesn't fix the root cause. Option B is wrong because manual intervention is not sustainable.

Option D is wrong because using a different primary key won't address schema changes.

Full explanation →
386
MCQeasy

An admin is setting up Einstein Article Recommendations. Which type of data is essential for the model to learn which articles are relevant?

A.Article publication dates
B.Article view events from users
C.User job titles
D.Article author names
AnswerB

View events are the primary input for collaborative filtering.

Why this answer

Einstein Article Recommendations uses a collaborative filtering model that learns article relevance from user interaction signals, specifically article view events. The model analyzes patterns of which articles users view together to identify related content, making view events the essential training data for generating recommendations.

Exam trap

Salesforce often tests the distinction between essential training data (user behavior signals like view events) and optional metadata (like publication dates or author names), leading candidates to mistakenly choose metadata that seems relevant but is not required for the collaborative filtering model to learn article relevance.

How to eliminate wrong answers

Option A is wrong because article publication dates are metadata that influence recency but are not used as primary training signals for collaborative filtering; the model learns relevance from user behavior, not timestamps. Option C is wrong because user job titles are demographic attributes that could be used for personalization but are not essential for the core recommendation model, which relies on interaction data like views. Option D is wrong because article author names are content metadata that do not provide the behavioral signals needed for the model to learn which articles are relevant to users.

Full explanation →
387
MCQmedium

A team is developing a chatbot for customer service. To ensure ethical AI, which practice should be incorporated?

A.Allow the chatbot to escalate to a human agent upon request.
B.Store all conversation data indefinitely for analysis.
C.Use a single data source for training to avoid inconsistency.
D.Design the chatbot to mimic human emotions perfectly.
AnswerA

Human escalation provides oversight and user control.

Why this answer

Option A is correct because human escalation supports accountability and user control. Option B is wrong because indefinite data storage violates privacy. Option C is wrong because mimicking emotions may deceive users.

Option D is wrong because a single data source may introduce bias.

Full explanation →
388
MCQmedium

A marketing manager wants to use Einstein Send Time Optimization. To generate personalized send time recommendations, which data does the model primarily rely on?

A.The individual contact's past email open and click behavior.
B.The aggregated engagement data of all contacts in the same time zone.
C.The industry benchmarks for optimal send times.
D.The sender's historical campaign performance by hour.
AnswerA

This is the core data used for personalized predictions.

Why this answer

Einstein Send Time Optimization (STO) uses a machine learning model that analyzes each individual contact's historical email engagement patterns—specifically their past open and click behavior—to predict the optimal send time unique to that contact. This personalized approach ensures that each recipient receives the email when they are most likely to engage, rather than relying on aggregate or rule-based heuristics.

Exam trap

Salesforce often tests the distinction between personalized (contact-level) and aggregated (cohort or sender-level) optimization, leading candidates to mistakenly choose time-zone or campaign-based options when the core requirement is individual behavioral modeling.

How to eliminate wrong answers

Option B is wrong because aggregated engagement data of all contacts in the same time zone ignores individual behavioral differences; Einstein STO builds a per-contact model, not a time-zone cohort model. Option C is wrong because industry benchmarks for optimal send times are generic averages and cannot account for the unique, learned patterns of each individual contact. Option D is wrong because the sender's historical campaign performance by hour reflects the sender's overall audience behavior, not the personalized, contact-level timing that Einstein STO requires.

Full explanation →
389
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

Option D is correct because 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.

Full explanation →
390
MCQmedium

Based on the exhibit, what does the accuracy of 0.85 indicate?

A.85% of the features are important for prediction.
B.85% of predictions that the opportunity will be won are correct.
C.85% of the model's predictions matched the actual outcomes.
D.85% of opportunities in the training data were won.
AnswerC

Accuracy measures overall correctness.

Why this answer

Accuracy is defined as the ratio of correctly predicted instances (both true positives and true negatives) to the total number of predictions. An accuracy of 0.85 means that 85% of the model's predictions (whether 'won' or 'lost') matched the actual outcomes in the dataset. This is a standard classification metric that evaluates overall correctness, not just one class.

Exam trap

Salesforce often tests the distinction between accuracy and precision, so the trap here is that candidates confuse 'accuracy' with 'precision' (the percentage of positive predictions that are correct) and incorrectly select Option B.

How to eliminate wrong answers

Option A is wrong because accuracy measures prediction correctness, not feature importance; feature importance is determined by techniques like permutation importance or SHAP values, not by the accuracy score. Option B is wrong because accuracy considers all predictions (both won and lost), not just the precision of 'won' predictions; 85% accuracy does not imply that 85% of 'won' predictions are correct—that would be precision. Option D is wrong because accuracy is computed on predictions versus actual outcomes, not on the distribution of the training data; the percentage of won opportunities in the training data is the class prior, not a performance metric.

Full explanation →
391
MCQeasy

A company wants to use Einstein Prediction Builder to predict customer churn. Which data preparation step is essential before building the model?

A.Ensure the data is in a Salesforce connected data source like Data Cloud.
B.Define the prediction objective and the target date field.
C.Create a formula field to calculate the churn probability.
D.Create a new custom object to store the prediction results.
AnswerB

The prediction objective (e.g., churn) is required to train the model.

Why this answer

Option B is correct because Einstein Prediction Builder requires you to define the prediction objective (e.g., 'Will this customer churn?') and specify the target date field that marks the event. This step is essential as it tells the model what to predict and over what time window, enabling the automated feature engineering and model training process.

Exam trap

Salesforce often tests the misconception that data must come from Data Cloud or that you need to pre-create storage objects, when in fact the core prerequisite is simply defining the prediction objective and target date field.

How to eliminate wrong answers

Option A is wrong because while Data Cloud is a supported data source, it is not mandatory; Einstein Prediction Builder can also use standard or custom objects directly in Salesforce. Option C is wrong because formula fields cannot be used to calculate churn probability; the model generates probability scores automatically after training, and you do not pre-compute them. Option D is wrong because prediction results are stored automatically in a standard Salesforce object (PredictionResult) or can be written to a field on the record; you do not need to create a custom object for storage.

Full explanation →
392
MCQmedium

A company notices that Einstein Prediction Builder predictions for 'Churn' are less accurate than expected. Which action should the administrator take first to improve model performance?

A.Enable field history tracking on all object fields used in the prediction.
B.Review the training data for missing values and ensure relevant fields are included in the model.
C.Change the prediction outcome to a different field to see if accuracy improves.
D.Retrain the model with the same data but increase the number of training iterations.
AnswerB

Data quality is fundamental; Einstein models rely on clean, relevant data.

Why this answer

Option B is correct because the first step in improving Einstein Prediction Builder model performance is to review the training data for missing values and ensure relevant fields are included. Missing values or irrelevant fields can introduce noise and bias, directly degrading predictive accuracy. Einstein Prediction Builder relies on high-quality, complete training data to learn meaningful patterns, so data quality issues must be addressed before any other tuning steps.

Exam trap

Salesforce often tests the misconception that retraining or tweaking model parameters is the first fix for poor accuracy, when in reality data quality review is the foundational step in any machine learning workflow.

How to eliminate wrong answers

Option A is wrong because enabling field history tracking on all object fields is unnecessary and can cause excessive data storage and performance overhead; field history tracking is used for auditing changes, not for improving model accuracy. Option C is wrong because changing the prediction outcome to a different field does not fix underlying data quality issues; it merely shifts the target variable without addressing why the current model is underperforming. Option D is wrong because retraining the model with the same data and increasing training iterations will not compensate for missing values or irrelevant fields; it can lead to overfitting on flawed data rather than improving generalization.

Full explanation →
393
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

Option D is correct because 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.

Full explanation →
394
Multi-Selecteasy

A company is developing an AI system to screen job applicants. Which TWO practices are essential for ethical AI in hiring?

Select 2 answers
A.Using all available data including demographic details
B.Auditing the model for bias against protected groups
C.Relying solely on AI for final decisions
D.Maximizing processing speed
E.Providing candidates with explanation of decisions
AnswersB, E

Bias auditing is crucial to ensure fairness.

Why this answer

Options B and D are essential: auditing for bias (B) and ensuring transparency (D). Option A is wrong because speed is not ethical. Option C is wrong as using all data may include biased data.

Option E is wrong because sole reliance on AI is unethical without human oversight.

Full explanation →
395
MCQhard

Based on the exhibit, what is the primary issue with this Einstein Bot conversation?

A.The bot lacks alternative ways to identify the customer.
B.The bot is confused about the user's intent.
C.The bot is repeating itself excessively.
D.The bot does not understand the initial intent.
AnswerA

The bot should offer alternatives like email lookup.

Why this answer

The exhibit shows the bot repeatedly asking for the customer's account number without offering alternative identification methods (e.g., email, phone number, or name). This is the primary issue because Einstein Bot's conversational design should include fallback paths to handle cases where the user cannot provide the requested information, ensuring a smooth user experience and reducing drop-offs.

Exam trap

Salesforce often tests the misconception that the primary issue is intent confusion or repetition, but the real trap is recognizing that the bot's inability to offer alternative identification methods is a design flaw in the dialog flow, not a failure of NLU or looping logic.

How to eliminate wrong answers

Option B is wrong because the bot correctly identifies the user's intent (e.g., 'I need help with my bill') and proceeds to gather account details, so there is no confusion about intent. Option C is wrong because the bot does not repeat itself excessively; it asks for the account number only once per turn, and the repetition is due to the user not providing it, not a loop error. Option D is wrong because the bot understands the initial intent (e.g., billing inquiry) and responds appropriately, so the issue is not a failure to understand intent but a lack of alternative identification methods.

Full explanation →
396
MCQmedium

A company uses an AI model to automate customer service responses. A customer receives an incorrect response that results in a financial loss. Who is primarily accountable for this error?

A.The AI model
B.The developer who built the model
C.The organization that deployed the AI
D.The customer who received the response
AnswerC

Organizations are accountable for the systems they deploy.

Why this answer

The correct answer is B because the organization deploying the AI is accountable for its outcomes. Option A is wrong because the AI itself is not accountable. Option C is wrong because the developer may share responsibility, but ultimate accountability lies with the organization.

Option D is wrong because the customer is not responsible.

Full explanation →
397
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

Option C is correct: 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.

Full explanation →
398
MCQhard

A bank uses Einstein Discovery to generate insights about loan approval decisions. After deployment, they notice the model denies loans to a higher percentage of applicants from a certain postal code. Which action should be taken to ensure responsible AI?

A.Ignore the discrepancy because postal code is not a protected attribute
B.Retrain the model using only recent loan data
C.Audit model outcomes for fairness across demographic groups and retrain if needed
D.Remove the postal code field from the model
AnswerC

Bias audit and mitigation is a standard responsible AI practice.

Why this answer

Option C is correct because responsible AI requires auditing model outcomes for fairness across demographic groups, even when the disparity correlates with a non-protected attribute like postal code. In Einstein Discovery, postal code can act as a proxy for protected attributes such as race or socioeconomic status, and ignoring this could lead to discriminatory lending practices. Auditing allows the team to detect and mitigate bias, and retraining with fairness constraints ensures the model aligns with ethical AI principles.

Exam trap

Salesforce often tests the misconception that removing a sensitive feature (like postal code) automatically eliminates bias, when in reality proxy features and correlated variables can still cause unfair outcomes.

How to eliminate wrong answers

Option A is wrong because ignoring the discrepancy is irresponsible; postal code can be a proxy for protected attributes (e.g., race or income), and model fairness must be evaluated even if the field itself is not protected. Option B is wrong because retraining on only recent loan data does not address the root cause of bias; it may even amplify existing disparities if recent data still reflects historical biases or sampling issues. Option D is wrong because simply removing the postal code field does not guarantee fairness; other correlated features (e.g., income, credit history) can still encode the same bias, and the model may still discriminate indirectly through proxy variables.

Full explanation →
399
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

Option C is correct because 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.

Full explanation →
400
MCQhard

A financial institution uses an AI model to approve credit. The model shows disparate impact against a protected group. Under Salesforce's ethical AI principles, what is the most appropriate action?

A.Discontinue the model immediately.
B.Increase the model's decision threshold for all applicants.
C.Publish the model's predictions publicly for scrutiny.
D.Apply a bias mitigation technique such as reweighing.
AnswerD

Bias mitigation techniques directly address disparate impact.

Why this answer

Option B is correct because bias mitigation techniques like reweighing address disparate impact while maintaining functionality. Option A is wrong because discontinuing the model may cause operational disruption and is not necessary if bias can be mitigated. Option C is wrong because adjusting the threshold may not address the root cause.

Option D is wrong because publicizing predictions may violate privacy.

Full explanation →
401
MCQmedium

To integrate external data into Salesforce for AI, which tool is recommended by Salesforce for building data pipelines?

A.Salesforce Connect
B.Data Export Service
C.Salesforce Data Pipelines
D.Apex Data Loader
AnswerC

Data Pipelines provides a visual interface to build and schedule data transformations for AI.

Why this answer

Option B is correct because Salesforce Data Pipelines is the recommended tool for creating and managing data integration workflows for AI and analytics. The other options are not designed for this purpose.

Full explanation →
402
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.

Full explanation →
403
MCQmedium

A company uses Salesforce Data Platform to store customer data. They want to use this data to train an AI model for lead scoring, but they are concerned about data quality. Which step should they take first to ensure the data is suitable for AI?

A.Profile the data to identify missing values, outliers, and inconsistencies
B.Immediately normalize all numerical features
C.Create a labeled dataset using historical lead outcomes
D.Set up a data pipeline to stream data in real-time
AnswerA

Profiling is the first step to assess data quality.

Why this answer

Profiling the data is the essential first step because it systematically identifies missing values, outliers, and inconsistencies that degrade model performance. Without this baseline assessment, any subsequent normalization or labeling would be applied to flawed data, leading to unreliable lead scoring predictions. Salesforce Data Platform supports profiling via tools like Einstein Analytics or Data Prep, which scan fields for nulls, range violations, and format errors.

Exam trap

Salesforce often tests the misconception that data preparation begins with feature engineering (like normalization) or pipeline setup, rather than with foundational data quality assessment through profiling.

How to eliminate wrong answers

Option B is wrong because normalizing numerical features is a preprocessing step that should only occur after data quality issues (like missing values or outliers) have been identified and resolved; applying normalization prematurely can amplify the impact of corrupt data. Option C is wrong because creating a labeled dataset is a critical step for supervised learning, but it assumes the raw data is already clean and consistent, which is not the case when data quality is a concern. Option D is wrong because setting up a real-time data pipeline addresses data velocity and freshness, not data quality; streaming dirty data into the pipeline would only propagate errors faster.

Full explanation →
404
MCQeasy

Refer to the exhibit. The prediction API returns a probability of 0.85 for the label 'High Value'. What does this value represent?

A.The likelihood that this lead will convert
B.The confidence score that this lead is 'High Value'
C.The F1 score of the model for this prediction
D.The model's accuracy on the training set
AnswerB

The score indicates how sure the model is about the predicted label.

Why this answer

The prediction API returns a probability of 0.85 for the label 'High Value'. In machine learning classification, this output represents the model's confidence score—the estimated probability that the input instance belongs to the specified class. It is not a direct measure of conversion likelihood, model accuracy, or F1 score; it is the raw posterior probability assigned by the model to the 'High Value' label.

Exam trap

Salesforce often tests the distinction between a model's per-instance confidence score and aggregate performance metrics like accuracy or F1 score, trapping candidates who confuse the output of a prediction API with evaluation metrics.

How to eliminate wrong answers

Option A is wrong because the probability 0.85 is the model's confidence that the lead belongs to the 'High Value' class, not a direct prediction of conversion likelihood—conversion is a separate business outcome that may depend on other factors. Option C is wrong because the F1 score is a model evaluation metric computed from precision and recall over a test set, not a per-prediction output from the API. Option D is wrong because the model's accuracy on the training set is a global performance metric, not a per-instance probability returned by the prediction API.

Full explanation →
405
MCQhard

An organization uses Salesforce Data Cloud to unify customer data from multiple sources. They want to ensure that data lineage is tracked for AI models. Which practice supports data lineage?

A.Use data partitioning to improve query performance.
B.Implement role-based access control on datasets.
C.Maintain metadata that records source, transformations, and dependencies.
D.Regularly run data profiling to check completeness.
AnswerC

Metadata enables lineage tracking.

Why this answer

Maintaining metadata that records source, transformations, and dependencies is the correct practice because data lineage for AI models requires a complete audit trail of where data originated, how it was transformed, and its dependencies. In Salesforce Data Cloud, this metadata is captured through the Data Catalog and Data Lineage feature, which tracks the flow of data from source objects through calculated insights and segments to AI model inputs, ensuring transparency and reproducibility.

Exam trap

Salesforce often tests the distinction between data management practices that improve performance or security versus those that specifically support auditability and traceability, leading candidates to confuse data partitioning or access control with lineage tracking.

How to eliminate wrong answers

Option A is wrong because data partitioning improves query performance by dividing data into smaller segments, but it does not track the origin, transformation steps, or dependencies of data, which are essential for lineage. Option B is wrong because role-based access control (RBAC) governs who can view or modify datasets, but it provides no record of data provenance or transformation history. Option D is wrong because data profiling checks completeness, accuracy, and consistency of data, but it does not capture the sequence of transformations or source-to-target mappings required for lineage.

Full explanation →
406
MCQhard

A team is labeling text data for a sentiment analysis model. To ensure consistency and quality, which practice should they prioritize?

A.Use a single expert labeler for all data.
B.Use majority voting among multiple labelers.
C.Label all data by a single expert labeler.
D.Allow each labeler to interpret guidelines freely.
AnswerB

Majority voting aggregates judgments, improving accuracy and consistency.

Why this answer

Majority voting among multiple labelers reduces individual bias and errors, improving label consistency and quality for training data. This approach is standard in supervised learning for sentiment analysis because it aggregates diverse judgments, leading to more reliable ground truth labels.

Exam trap

Salesforce often tests the misconception that a single expert labeler guarantees higher quality, when in fact multiple labelers with majority voting reduce bias and improve reliability for training data.

How to eliminate wrong answers

Option A is wrong because using a single expert labeler introduces individual bias and lacks error checking, which can degrade model performance due to inconsistent or subjective labels. Option C is wrong because labeling all data by a single expert labeler is identical to Option A and suffers from the same lack of consensus and quality assurance. Option D is wrong because allowing each labeler to interpret guidelines freely leads to high inter-labeler variability, undermining consistency and making the dataset unreliable for training a robust model.

Full explanation →
407
MCQmedium

Refer to the exhibit. What effect does this masking policy have on the data used for training an Einstein model?

A.Only SSN is masked.
B.SSN and CreditCard fields are encrypted.
C.SSN and CreditCard fields are completely removed from training data.
D.SSN and CreditCard fields are partially masked, showing only the last four characters.
AnswerD

Explicitly defined by showLastFour and maskingType partial.

Why this answer

The masking policy in Einstein applies a partial mask to sensitive fields like SSN and CreditCard, showing only the last four characters while obscuring the rest. This ensures that the data used for training retains its structural utility for model learning without exposing full sensitive values, which is why option D is correct.

Exam trap

Salesforce often tests the distinction between masking, encryption, and removal, where candidates mistakenly think masking is equivalent to encryption or complete deletion, but masking specifically preserves partial data for model training while hiding sensitive details.

How to eliminate wrong answers

Option A is wrong because the masking policy applies to both SSN and CreditCard fields, not just SSN, as indicated by the exhibit showing both fields being masked. Option B is wrong because masking is not encryption; encryption transforms data into a ciphertext that requires a key to reverse, whereas masking irreversibly obscures parts of the data for privacy. Option C is wrong because the policy does not completely remove the fields; it partially masks them, leaving the last four characters visible for training purposes.

Full explanation →
408
Multi-Selectmedium

Which TWO data preparation steps are critical for ensuring high-quality training data?

Select 2 answers
A.Increasing dataset size by adding noise.
B.Removing duplicate records.
C.Normalizing all features.
D.Handling missing values appropriately.
E.Using only labeled data.
AnswersB, D

Duplicates can overrepresent certain patterns and skew model training.

Why this answer

Option B is correct because duplicate records in a dataset can cause the model to overfit to repeated patterns, biasing the learned distribution and reducing generalization. Removing duplicates ensures each data point contributes equally to training, which is essential for robust model performance.

Exam trap

Salesforce often tests the distinction between data preparation steps that ensure data quality (like removing duplicates and handling missing values) versus optional preprocessing or augmentation techniques, leading candidates to mistakenly select normalization or noise addition as critical steps.

Full explanation →
409
MCQeasy

A company plans to train an AI model using data from Salesforce CRM and an external marketing automation platform. What is the first step to unify these data sources in Data Cloud?

A.Define a Data Model that maps fields from both sources to a unified customer object
B.Create two separate Data Streams to bring data in
C.Build a Calculated Insight to merge the data
D.Set up a Data Transformation to blend the sources
AnswerA

Unifies the schema before ingestion.

Why this answer

Option A is correct because creating a data model that maps fields from both sources to a common object ensures consistency. Option B is wrong because data streams come after the model. Option C is wrong because Calculated Insights are for aggregations.

Option D is wrong because data transformations are applied later.

Full explanation →
410
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

Option D is correct because 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.

Full explanation →
411
MCQmedium

An HR department uses an AI tool to screen resumes for a software engineering position. The tool was trained on resumes of past successful hires, who were predominantly male. The tool has been in use for three months, during which only 10% of candidates shortlisted for interviews are female, even though 40% of applicants are female. The hiring managers are satisfied with the quality of candidates shortlisted, as most perform well in interviews. However, the company's diversity and inclusion officer raises an ethical concern. What should the company do to address this bias?

A.Retrain the model with a balanced dataset that includes more female candidates and remove gender-correlated features.
B.Use the AI tool for initial screening but allow candidates to appeal the decision.
C.Continue using the current tool since it selects high-quality candidates.
D.Manually review all resumes without using the AI tool.
AnswerA

This reduces bias by ensuring the model learns from diverse examples and avoids proxy discrimination.

Why this answer

Option B is correct because retraining with a balanced dataset and using gender-blind features directly addresses the source of bias. Option A perpetuates the bias. Option C is too labor-intensive for high volume.

Option D still relies on the biased model for initial screening.

Full explanation →
412
MCQhard

A system administrator receives an error when running a Data Cloud data transform: 'Row-level security settings are preventing access to the source data.' The admin has appropriate permissions. What is the most likely cause?

A.The target object has field-level security.
B.The data stream is scheduled during maintenance.
C.The data transform is set to run as the admin's default user.
D.The source object has sharing rules that restrict access for the data transform's running user.
AnswerD

Row-level security is about sharing; the running user may not see all rows.

Why this answer

Option D is correct because Data Cloud data transforms run under a specific running user context, and row-level security (RLS) settings on the source object can restrict that user's access to rows, even if the admin has broad permissions. The error indicates that the running user lacks visibility to certain source data rows due to sharing rules or RLS policies, which is a common cause of this specific error message.

Exam trap

Salesforce often tests the distinction between row-level security and field-level security, and candidates mistakenly choose field-level security (Option A) because they confuse the two concepts, not realizing the error message explicitly points to row-level restrictions.

How to eliminate wrong answers

Option A is wrong because field-level security (FLS) controls access to fields, not rows, and the error explicitly mentions 'row-level security,' not field-level. Option B is wrong because scheduled maintenance would typically cause a different error (e.g., 'service unavailable' or timeout), not a row-level security access denial. Option C is wrong because running as the admin's default user would inherit the admin's permissions, which should have access; the error states the admin has appropriate permissions, so the issue is with the running user context, not the default user setting.

Full explanation →
413
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.

Full explanation →
414
Multi-Selecteasy

A company is deploying Einstein Vision for product quality inspection. To ensure ethical use, which TWO practices should they adopt? (Choose two.)

Select 2 answers
A.Test the model for bias across different product types and lighting conditions
B.Keep the model's decision-making process proprietary to protect intellectual property
C.Deploy the model without human oversight to maximize efficiency
D.Provide clear documentation on the model's limitations and expected accuracy
E.Only use training images from a single supplier to maintain consistency
AnswersA, D

Bias testing ensures fair performance.

Why this answer

Option A (Test the model for bias across different product types) is correct because bias testing is essential. Option C (Provide clear documentation on the model's limitations) is correct for transparency. Option B (Use the model without human oversight) violates accountability.

Option D (Only use images from one supplier) may introduce bias. Option E (Keep the model's decisions secret) violates transparency.

Full explanation →
415
MCQhard

An organization is preparing data for Einstein Next Best Action. They have multiple action types (discounts, product suggestions, content). Which data model approach best ensures accurate recommendations?

A.Train a separate model per customer segment and then merge.
B.Create a separate model for each action type and combine results manually.
C.Build an ensemble of models and average their outputs.
D.Use a single model that includes all action types in the training data.
AnswerD

A unified model captures interactions between actions, leading to better optimization of the next best action.

Why this answer

Option D is correct because Einstein Next Best Action is designed to learn from all action types simultaneously within a single model. By including all action types (discounts, product suggestions, content) in the training data, the model can capture cross-action patterns and relative effectiveness, leading to more accurate and contextually relevant recommendations. A unified model avoids fragmentation and ensures consistent scoring across actions.

Exam trap

Salesforce often tests the misconception that separate models per action or segment improve accuracy, when in fact Einstein Next Best Action requires a single unified model to learn cross-action patterns and deliver coherent recommendations.

How to eliminate wrong answers

Option A is wrong because training a separate model per customer segment and merging results introduces fragmentation and ignores cross-segment patterns, reducing the model's ability to generalize and leading to inconsistent recommendations. Option B is wrong because creating a separate model for each action type and manually combining results loses the interdependencies between actions, such as when a discount makes a product suggestion more effective, and adds unnecessary complexity without leveraging Einstein's built-in multi-action support. Option C is wrong because building an ensemble of models and averaging outputs does not align with Einstein Next Best Action's architecture, which expects a single unified model to optimize across all actions; averaging can dilute the signal from specific action types and reduce recommendation precision.

Full explanation →
416
MCQmedium

A global e-commerce company deploys Einstein Bots in multiple countries. The bot uses natural language processing to handle customer returns. In one region, customers frequently complain that the bot does not understand their local dialect and incorrectly rejects valid returns. The company wants to maintain consistent customer experience while respecting regional diversity. The bot's language model was trained mainly on English data from the US and UK. The AI ethics board is concerned about fairness and transparency. They consider four options: (A) use a single, centrally-trained model with fallback to human agents for non-English queries, (B) deploy separate models fine-tuned on each dialect but with centralized monitoring, (C) disable the bot in regions with dialect issues, (D) use a translation layer to convert all inputs to English before processing. What is the best ethical approach?

A.Use a single centrally-trained model with fallback to human agents for non-English queries
B.Use a translation layer to convert all inputs to English before processing
C.Disable the bot in regions with dialect issues
D.Deploy separate models fine-tuned on each dialect with centralized monitoring
AnswerD

Fine-tuning respects linguistic diversity and central monitoring ensures consistent ethics.

Why this answer

Option B is correct because fine-tuning on local dialects improves accuracy and fairness, while centralized monitoring ensures oversight. Option A is wrong: fallback to human agents is good but still may cause delays and dissatisfaction; the model itself is not inclusive. Option C is wrong: disabling the bot denies service and is not inclusive.

Option D is wrong: translation may lose nuance and introduce errors.

Full explanation →
417
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.

Full explanation →
418
MCQeasy

A sales manager wants to use Einstein Activity Capture to log emails automatically. Which prerequisite must be met?

A.The org must be on Enterprise Edition or higher.
B.The user's email must be hosted on a supported platform (Gmail, Outlook).
C.The user must have an Einstein AI license.
D.The user must manually enable email logging in personal settings.
AnswerB

Einstein Activity Capture integrates with supported email providers.

Why this answer

Einstein Activity Capture requires that user emails be hosted on a supported platform (Gmail or Outlook/Exchange) because the feature uses server-side synchronization via APIs (Google Workspace APIs or Microsoft Graph) to automatically log emails and events into Salesforce. Without a supported email host, the service cannot connect to the mail server to capture activity data.

Exam trap

The trap here is that candidates often assume Einstein Activity Capture requires a higher edition (Enterprise+) or an AI license, when in fact the critical prerequisite is a supported email platform, not the edition or an add-on license.

How to eliminate wrong answers

Option A is wrong because Einstein Activity Capture is available on Professional, Enterprise, and Unlimited Editions, not exclusively on Enterprise Edition or higher. Option C is wrong because Einstein Activity Capture does not require an Einstein AI license; it is included with Sales Cloud or Service Cloud licenses that support the feature. Option D is wrong because email logging in Einstein Activity Capture is automatic once configured by an administrator; users do not need to manually enable it in personal settings.

Full explanation →
419
Multi-Selecteasy

Which THREE are core principles in Salesforce's AI ethics framework?

Select 3 answers
A.Accuracy
B.Fairness
C.Profitability
D.Transparency
E.Privacy
AnswersB, D, E

Core principle.

Why this answer

Option A is correct because privacy is a core principle. Option C is correct because fairness is a core principle. Option D is correct because transparency is a core principle.

Option B is wrong while important, accuracy is not explicitly a core ethical principle in Salesforce's framework; they focus on fairness, transparency, accountability, privacy, and safety. Option E is wrong profitability is not an ethical principle.

Full explanation →
420
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.

Full explanation →
421
MCQeasy

Refer to the exhibit. A developer receives this fairness check error. What is the most likely cause?

A.The model has higher false positive and false negative rates for Group B.
B.The error is due to insufficient training data for Group A.
C.The model is overfitting.
D.The recommendation suggests reweighting, so the model is already fair.
AnswerA

The higher rates for Group B indicate bias.

Why this answer

Option B is correct because the exhibit clearly shows significantly higher false positive and false negative rates for Group B, indicating the model treats groups differently. Option A is wrong because overfitting is not indicated by these metrics. Option C is wrong because the recommendation is to fix the issue, not that the model is already fair.

Option D is wrong because Group A has lower rates, suggesting Group B's data may be insufficient, but the cause is the disparity itself.

Full explanation →
422
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.

Full explanation →
423
MCQhard

A company uses Salesforce Data Cloud to unify customer data from multiple sources for AI model training. After adding a new data source, model performance degrades significantly. What is the most likely cause?

A.Insufficient compute resources
B.Data labeling errors
C.Data schema mismatch
D.Data duplication from overlapping sources
AnswerD

Duplication introduces bias and degrades performance.

Why this answer

Option A is correct because data duplication due to overlapping records from multiple sources can bias the model. Option B is wrong because schema mismatch would cause load errors, not just performance degradation. Option C is wrong because compute issues would affect all models.

Option D is wrong because data labeling errors would affect the training process, not the data unification step.

Full explanation →
424
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.

Full explanation →
425
Multi-Selecteasy

A company wants to use Einstein Relationship Intelligence to analyze email and calendar data for opportunity insights. Which two conditions must be met? (Select two answers.)

Select 2 answers
A.Exchange or Gmail must be used for email.
B.The feature is only available in Unlimited Edition.
C.Users must have the Einstein Relationship Intelligence permission set.
D.Einstein Relationship Intelligence must be enabled in the org.
E.All users must opt in to email logging.
AnswersC, D

The permission set allows users to view relationship insights.

Why this answer

Option C is correct because the Einstein Relationship Intelligence permission set grants users access to the feature's insights and data processing capabilities. Without this permission set, users cannot view or interact with Einstein Relationship Intelligence data, even if the feature is enabled at the org level.

Exam trap

Salesforce often tests the distinction between org-level feature enablement and user-level permission assignment, leading candidates to select only one of these two required conditions.

Full explanation →
426
MCQeasy

A company wants to use Einstein Article Recommendations to suggest knowledge articles to support agents. What is a prerequisite for this feature?

A.Articles must be of a specific type, such as FAQ.
B.The org must be enabled for Einstein features.
C.A case must be open for the recommendation to appear.
D.Knowledge articles must be created and published.
AnswerD

Articles must exist to be recommended.

Why this answer

Einstein Article Recommendations requires that knowledge articles are created and published in the Salesforce Knowledge base. The feature uses natural language processing (NLP) to match the context of a case or conversation with published articles, so unpublished or draft articles cannot be recommended. Without published articles, the AI model has no content to analyze or suggest.

Exam trap

Salesforce often tests the distinction between general Einstein enablement and feature-specific prerequisites, so candidates mistakenly select Option B thinking that enabling Einstein is the only requirement, when in fact published articles are the critical prerequisite for Article Recommendations to function.

How to eliminate wrong answers

Option A is wrong because Einstein Article Recommendations does not require articles to be of a specific type like FAQ; it works with any standard or custom article type defined in Salesforce Knowledge. Option B is wrong because while Einstein features generally require the org to be enabled for Einstein, this is a platform-level prerequisite for all Einstein services, not a specific prerequisite for Article Recommendations—the question asks for a prerequisite specific to this feature. Option C is wrong because a case does not need to be open for recommendations to appear; Einstein can also suggest articles in other contexts such as Chat, Email-to-Case, or even in the Knowledge tab without an open case.

Full explanation →
427
MCQeasy

Refer to the exhibit. A data transformation configuration is shown. Which of the following describes the outcome of applying this transformation?

A.The transformation is invalid because one-hot encoding cannot be combined with scaling.
B.Only 'color' is transformed; 'price' and 'weight' are unchanged.
C.'color' is one-hot encoded into multiple binary columns; 'price' and 'weight' are standardized to have mean 0 and variance 1.
D.'color' is scaled to [0,1] and 'price', 'weight' are one-hot encoded.
AnswerC

Correct interpretation of the config.

Why this answer

Option C is correct because the transformation configuration applies a one-hot encoder to the 'color' categorical column, creating multiple binary columns, and applies a standard scaler to the 'price' and 'weight' numerical columns, centering them to mean 0 and scaling to unit variance. This is a common preprocessing pipeline that handles mixed data types appropriately.

Exam trap

Salesforce often tests the ability to distinguish which transformation applies to which column type, trapping candidates who confuse scaling with encoding or assume that different transformations cannot coexist in a single pipeline.

How to eliminate wrong answers

Option A is wrong because one-hot encoding and scaling can be combined in a single transformation pipeline; they are applied to different columns (categorical vs. numerical) and are not mutually exclusive. Option B is wrong because the transformation explicitly applies a standard scaler to 'price' and 'weight', so they are not unchanged; they are standardized. Option D is wrong because it reverses the operations: 'color' is one-hot encoded, not scaled to [0,1], and 'price' and 'weight' are standardized, not one-hot encoded.

Full explanation →
428
MCQhard

A global manufacturing company uses Sales Cloud and has implemented Einstein Opportunity Scoring to prioritize deals. The scoring model was trained on historical data and initially performed well. Over the past month, the scores have become less accurate, with many high-scoring opportunities not closing and some low-scoring ones closing. The admin notices that the sales team has been using a new discounting strategy that heavily influences deal outcomes. The admin wants to improve model performance without manual intervention. Which action should the admin take?

A.Manually adjust the field weights for discount-related fields in the model.
B.Retrain the Einstein Opportunity Scoring model with the latest opportunity data including discount information.
C.Run a data quality report to identify and clean missing discount data.
D.Create a custom field for discount percentage and add it to the model.
AnswerB

Retraining incorporates new patterns.

Why this answer

Option B is correct because retraining the Einstein Opportunity Scoring model with the latest opportunity data, including discount information, allows the machine learning model to automatically learn the new patterns introduced by the sales team's discounting strategy. This aligns with the AI Associate principle that models must be retrained on current data to maintain accuracy when business processes change, without requiring manual intervention.

Exam trap

Salesforce often tests the misconception that adding a field or cleaning data alone will improve model performance, when in fact the model must be retrained to incorporate the new data and learn the changed relationships.

How to eliminate wrong answers

Option A is wrong because manually adjusting field weights contradicts the 'without manual intervention' requirement and undermines the automated machine learning approach of Einstein Opportunity Scoring, which learns weights from data. Option C is wrong because running a data quality report to clean missing discount data addresses data completeness but does not cause the model to learn the new discounting strategy's impact on deal outcomes; the model still needs retraining to incorporate the changed behavior. Option D is wrong because creating a custom field for discount percentage and adding it to the model is a prerequisite step, but it alone does not improve model performance; the model must be retrained with the new field and latest data to adjust its scoring logic.

Full explanation →
429
MCQeasy

Refer to the exhibit. What data quality issue does the exhibit reveal?

A.The Summer Sale campaign has duplicate records.
B.The Fall Clearance campaign has no response data.
C.The query syntax is incorrect.
D.The data is not normalized.
AnswerB

NonNullResponse is 0, meaning all responses are null.

Why this answer

The Fall Clearance campaign has zero non-null responses, indicating all response data is missing for that campaign.

Full explanation →
430
MCQhard

Refer to the exhibit. An admin runs two queries on the Lead object. Both include the custom field Score__c used by Einstein Lead Scoring. The second query is significantly slower. What is the most likely cause?

A.The query is not using a selective filter on Score__c
B.The query uses a date function that is not selective
C.The Score__c field is indexed
D.The Einstein Lead Scoring job is running simultaneously
AnswerA

Large result set without index on Score__c causes slow performance.

Why this answer

Option A is correct because Einstein Lead Scoring uses the Score__c field to store lead scores, and the field is not indexed by default. When a query filters on Score__c without an index, it forces a full table scan on the Lead object, which becomes significantly slower as the number of leads grows. The second query likely includes a non-selective filter on Score__c (e.g., a range or inequality), which cannot leverage any existing index and thus degrades performance.

Exam trap

Salesforce often tests the misconception that Einstein features automatically index their underlying fields, when in fact custom fields like Score__c are not indexed by default and require manual indexing for query performance.

How to eliminate wrong answers

Option B is wrong because the question explicitly states both queries include the custom field Score__c, and there is no mention of a date function in either query; the performance difference is due to the filter on Score__c, not a date function. Option C is wrong because Score__c is a custom field used by Einstein Lead Scoring and is not automatically indexed; if it were indexed, the second query would likely be faster, not slower. Option D is wrong because Einstein Lead Scoring jobs run asynchronously in the background and do not directly impact the performance of individual SOQL queries; the slowdown is due to query selectivity, not concurrent job execution.

Full explanation →
431
MCQmedium

An admin is troubleshooting Einstein Sentiment. The model returns high confidence but wrong sentiment (e.g., positive reviews labeled negative). What is the most likely issue?

A.The model was not retrained after the last data load.
B.The training data contains predominantly neutral examples.
C.The training data has incorrect labels for sentiment.
D.The field mapping for the sentiment field is incorrect.
AnswerC

Garbage in, garbage out: mislabeled training data leads to confident but incorrect classifications.

Why this answer

Option C is correct because if the training data contains incorrect labels for sentiment, the model learns from erroneous ground truth, leading to high confidence in wrong predictions. In Einstein Sentiment, the model's accuracy depends directly on the quality and correctness of the labeled training data; mislabeled examples cause the classifier to associate features with the wrong sentiment class, resulting in confident but incorrect outputs.

Exam trap

Salesforce often tests the concept that high confidence does not imply high accuracy; candidates mistakenly assume retraining or data volume issues are the root cause, rather than recognizing that garbage-in (incorrect labels) leads to garbage-out (confident wrong predictions).

How to eliminate wrong answers

Option A is wrong because retraining after a data load does not fix incorrect labels; it would only reinforce the existing mislabeled patterns. Option B is wrong because predominantly neutral examples would bias the model toward neutral predictions, not cause high-confidence wrong sentiment (e.g., positive labeled as negative). Option D is wrong because incorrect field mapping would typically result in missing or misaligned data, not high-confidence misclassification of sentiment; the model would fail to train or predict altogether.

Full explanation →
432
MCQhard

A company wants to use Einstein Next Best Action but needs to ensure data privacy. What is the required step for anonymizing customer data in Data Pipelines?

A.Create a sandbox with scrambled data
B.Rely on Einstein's built-in anonymization
C.Use Permission Sets to hide fields from users
D.Use the Data Mask transformation in Data Pipelines
AnswerD

Data Mask can replace sensitive values with anonymized data at the pipeline level.

Why this answer

Option A is correct because Data Pipelines includes a Data Mask transformation that can anonymize PII fields. Permission Sets control access but do not anonymize; sandbox scrambling is for testing only; Einstein does not automatically anonymize data.

Full explanation →
433
MCQeasy

A Salesforce admin wants to use Einstein Prediction Builder to predict customer churn. What ethical consideration is most important?

A.Cost of implementation
B.Model accuracy
C.Data privacy
D.Transparency of predictions
AnswerC

Protecting customer data privacy is a core ethical requirement.

Why this answer

Option C is correct because data privacy is paramount when using customer data for predictions. Option A is wrong while accuracy is important, but privacy is a foundational ethical concern. Option B is wrong because transparency is important but not as fundamental as privacy in this context.

Option D is wrong because cost is a business, not ethical, concern.

Full explanation →
434
MCQeasy

Which data type is most commonly used for image recognition AI models?

A.Unstructured data
B.Structured data
C.Time-series data
D.Semi-structured data
AnswerA

Images are unstructured data.

Why this answer

Option D is correct because image recognition primarily uses unstructured data (pixel values). Option A is wrong because structured data (tables) is not suitable for images. Option B is wrong because semi-structured data (like JSON) is not typical.

Option C is wrong because time-series data is for sequential measurements.

Full explanation →
435
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

Option B is correct because 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.

Full explanation →
436
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

Option A is correct because 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.

Full explanation →
437
MCQmedium

A Salesforce admin is troubleshooting an Einstein Prediction Builder model that is not generating predictions. The model was created with a custom object 'Feedback__c'. The admin notices that the model's data source includes records with status 'In Progress' and 'Closed'. What is the most likely cause of the model not generating predictions?

A.The data source is not refreshed daily
B.The object is a custom object
C.The outcome field has more than two unique values
D.The object has fewer than 1000 records
AnswerC

Einstein Prediction Builder requires a binary outcome; status with 'In Progress' creates three distinct values.

Why this answer

Option C is correct because Einstein Prediction Builder requires the outcome field to have exactly two unique values (binary classification). The presence of 'In Progress' and 'Closed' as statuses suggests the outcome field likely contains more than two values, which violates this requirement and prevents the model from generating predictions.

Exam trap

The trap here is that candidates often assume any data quality issue (like stale data or record count) is the cause, but the specific requirement for binary outcome fields in Einstein Prediction Builder is a precise constraint that directly blocks prediction generation.

How to eliminate wrong answers

Option A is wrong because the data source refresh frequency does not affect the model's ability to generate predictions; it only impacts data freshness. Option B is wrong because custom objects are fully supported by Einstein Prediction Builder, and using a custom object does not prevent prediction generation. Option D is wrong because while a minimum of 1000 records is recommended for model training, having fewer records would cause a training failure, not a prediction generation failure after the model is built.

Full explanation →
438
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.

Full explanation →
439
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).

Full explanation →
440
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

Option C is correct because 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.

Full explanation →
441
MCQhard

A healthcare organization uses Salesforce to manage patient records. They want to deploy an AI system that predicts patient readmission risk. Which practice BEST ensures ethical use of patient data?

A.Use the model to deny high-risk patients coverage
B.Focus solely on model accuracy regardless of data source
C.Obtain explicit patient consent and anonymize data for training
D.Deploy a third-party AI tool without reviewing its data practices
AnswerC

Consent and anonymization protect patient privacy and comply with regulations.

Why this answer

Option B is correct because patient consent and data anonymization are key to ethical AI in healthcare. Option A is wrong because predictions may still lead to discrimination. Option C is wrong because accuracy alone doesn't guarantee ethical use.

Option D is wrong as third-party tools may not respect privacy laws.

Full explanation →
442
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.

Full explanation →
443
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

Option B is correct because 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.

Full explanation →
444
MCQmedium

Refer to the exhibit. An AI model audit shows performance differences across demographic groups. Which ethical concern is most critical?

A.Privacy: the data includes sensitive attributes
B.Accountability: the audit was not independent
C.Transparency: the model's overall accuracy is too low
D.Fairness: the model performs worse for the minority group
AnswerD

Biased performance across groups is a fairness issue.

Why this answer

Option A is correct: The significant disparity in accuracy and error rates indicates bias, which violates fairness. Option B is wrong because overall accuracy is high, but group fairness is lacking. Option C is wrong because accountability is about responsibility, but the exhibit directly shows unfairness.

Option D is wrong because privacy is not directly indicated.

Full explanation →
445
MCQmedium

A service manager wants to use Einstein Case Classification to automatically categorize incoming cases. What is a prerequisite for training the model?

A.A minimum of 100 open cases with categories assigned.
B.The cases must have been created within the last 30 days.
C.At least 500 closed cases with the correct category field populated.
D.Users must enable Einstein Case Classification in their personal settings.
AnswerC

This is the minimum requirement for model training.

Why this answer

Einstein Case Classification uses supervised machine learning to automatically categorize cases. The model learns from historical data, so it requires a sufficient number of closed cases (at least 500) with the correct category field populated to train effectively. Open cases are not used because the model needs confirmed outcomes to learn from.

Exam trap

Salesforce often tests the distinction between open and closed cases, and candidates mistakenly think open cases can be used for training because they are more recent or readily available, but the model requires confirmed historical data from closed cases.

How to eliminate wrong answers

Option A is wrong because the model requires closed cases, not open cases, as open cases lack the final category assignment needed for supervised learning. Option B is wrong because there is no 30-day creation window requirement; the model can use historical cases from any time period as long as they are closed and categorized. Option D is wrong because Einstein Case Classification is enabled at the system or profile level by an administrator, not in individual user personal settings.

Full explanation →
446
MCQhard

A healthcare AI model uses patient data. The legal team requires that all data used for training be de-identified according to HIPAA Safe Harbor method. Which data handling process satisfies this?

A.Remove all 18 HIPAA identifiers from each record.
B.Generate synthetic data that mimics patient records.
C.Remove patient names and replace with IDs.
D.Anonymize data by aggregating into groups of 10 or more.
AnswerA

Safe Harbor method requires removal of all listed identifiers.

Why this answer

Option A is correct because the HIPAA Safe Harbor method specifically requires the removal of all 18 identifiers listed in the HIPAA Privacy Rule from each patient record. This includes direct identifiers like names, addresses, and Social Security numbers, as well as indirect identifiers such as dates and geographic subdivisions. By removing these 18 identifiers, the data is considered de-identified and no longer subject to HIPAA restrictions, allowing it to be used for AI training.

Exam trap

The trap here is that candidates often confuse de-identification with anonymization or pseudonymization, assuming that removing just names or aggregating data is sufficient, but Cisco tests the specific requirement of removing all 18 HIPAA identifiers under the Safe Harbor method.

How to eliminate wrong answers

Option B is wrong because generating synthetic data that mimics patient records does not satisfy the HIPAA Safe Harbor method; while synthetic data can be useful, it is not a recognized de-identification method under HIPAA Safe Harbor, which requires the removal of specific identifiers from actual data. Option C is wrong because removing patient names and replacing them with IDs alone does not meet the Safe Harbor standard, as it still leaves 17 other identifiers (e.g., dates, ZIP codes) that must be removed. Option D is wrong because anonymizing data by aggregating into groups of 10 or more is not a HIPAA Safe Harbor method; aggregation may reduce re-identification risk but does not guarantee removal of all 18 identifiers, and Safe Harbor requires explicit removal of those identifiers, not statistical aggregation.

Full explanation →
447
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

Option B is correct because 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.

Full explanation →
448
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 correct answer is C because 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.

Full explanation →
449
MCQmedium

A company wants to use customer purchase history to train a recommendation model. Which action is essential to comply with data privacy regulations?

A.Use only publicly available data.
B.Ignore regulations because data is internal.
C.Obtain explicit consent from customers.
D.Anonymize the data after training.
AnswerC

Explicit consent is legally required for processing personal data for AI training.

Why this answer

Option C is correct because data privacy regulations such as GDPR and CCPA require a lawful basis for processing personal data, and explicit consent is a primary lawful basis when using customer purchase history for training a recommendation model. Without obtaining explicit consent, the company would be processing personal data without a valid legal ground, violating regulations that mandate transparency and user control over their data.

Exam trap

The trap here is that candidates often assume internal data is exempt from privacy regulations or that anonymization after training retroactively fixes compliance, but regulations require a lawful basis before any processing begins.

How to eliminate wrong answers

Option A is wrong because using only publicly available data does not guarantee compliance; the data may still contain personal information subject to privacy regulations, and the model could inadvertently infer private attributes from public data. Option B is wrong because internal data is not exempt from data privacy regulations; laws like GDPR apply to any processing of personal data regardless of whether it is internal or external. Option D is wrong because anonymizing data after training does not address the requirement for a lawful basis at the time of collection and processing; the model may have already learned patterns from identifiable data, and retroactive anonymization does not cure the initial lack of consent.

Full explanation →
450
MCQmedium

A company wants to integrate external customer behavior data into Salesforce to enhance AI predictions. Which Salesforce Data Cloud feature is specifically designed to ingest and map external data?

A.Apex triggers
B.Data Streams
C.Einstein Studio
D.Flow Builder
AnswerB

Data Streams ingest external data into Data Cloud.

Why this answer

Option D is correct because Data Streams in Data Cloud are used to bring in data from various external sources. Option A is wrong because Flow Builder is for automation. Option B is wrong because Apex triggers are custom code.

Option C is wrong because Einstein Studio is for building AI models, not data ingestion.

Full explanation →
Page 5

Page 6 of 7

Page 7

All pages

1234567

Practice AI Associate by domain

Target a specific domain to shore up weak areas.

AI FundamentalsAI Capabilities in CRMEthical Considerations of AIData for AI
See all domains with question counts →