AI Associate Practice Test 11 — 15 Questions

Question 1

A data analyst uses Einstein Discovery to analyze a dataset and receives a story that includes a waterfall chart and improvement suggestions. The analyst wants to share the insights with business users who don't have access to Einstein Discovery. Which three methods can they use to share the results?

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

Accepted Answer

Export the story as a PDF. Einstein Discovery allows users to export a story as a PDF, which can then be shared with business users who lack direct access to Einstein Discovery. This method provides a static, portable snapshot of the insights, including the waterfall chart and improvement suggestions, without requiring the recipients to have any Salesforce or Einstein licenses.

Answer

Send an email with the raw data

Answer

Create a custom mobile app

Question 2

Refer to the exhibit. A Salesforce CLI output shows the status of Einstein models in the org. Which model should the administrator investigate first?

Accepted Answer

Case_Escalation because it has an Error status.. An Einstein model with an 'Error' status indicates a critical failure that prevents the model from generating predictions or scoring records. This requires immediate investigation to restore functionality, as the model is non-operational and may impact business processes relying on its output.

Answer

Lead_Score_Model because its accuracy is lower than expected.

Answer

Campaign_Response because it has the oldest training date.

Answer

Opportunity_Forecast because it is still training.

Question 3

A sales manager wants to automatically surface important customer emails that require immediate attention. Which Einstein feature should they use?

Accepted Answer

Einstein Email Insights. Einstein Email Insights is the correct feature because it uses natural language processing (NLP) to analyze email content and automatically surface high-priority emails that require immediate attention. This directly addresses the sales manager's need to identify important customer emails without manual sorting.

Answer

Einstein Activity Capture

Answer

Einstein Lead Scoring

Answer

Einstein Opportunity Scoring

Question 4

A marketing director wants to use Einstein Engagement Scoring to prioritize leads. She has enabled Einstein and assigned the permission set to users. However, the Engagement Score field is not visible on any lead record. The admin checked the field-level security and it is visible to all profiles. What should the admin do next?

Accepted Answer

Add the Engagement Score field to the lead page layout.. The Engagement Score field is a standard field that must be added to the lead page layout to be visible on the record. Even though field-level security grants access, the field will not appear on the record detail page unless it is explicitly placed on the page layout. This is a common layout-level visibility requirement in Salesforce.

Answer

Verify that there are at least 500 leads with activity in the last 30 days.

Answer

Run the 'Calculate Einstein Engagement Scores' scheduled job.

Answer

Wait 24 hours for the model to train.

Question 5

A sales rep wants to generate personalized email drafts to send to prospects based on opportunity data. Which feature should they use within Salesforce?

Accepted Answer

Sales GPT. Sales GPT is the correct feature because it uses generative AI to create personalized email drafts based on opportunity data, such as deal stage, product interest, and customer interactions. It leverages large language models to generate context-aware content directly within Salesforce, enabling the sales rep to quickly produce tailored communications without manual effort.

Answer

Einstein Conversation Insights

Answer

Einstein Opportunity Scoring

Answer

Einstein Email Insights

Question 6

Which type of machine learning is used when a model is trained on historical sales data that includes both input features and the known outcome (e.g., closed won/lost) to predict whether a new lead will convert?

Accepted Answer

Supervised learning. Supervised learning is correct because the training dataset contains labeled examples — historical sales records where each row pairs input features (lead attributes) with a known outcome (closed won/lost). The model learns a mapping function from features to labels, then applies it to new, unlabeled leads to predict conversion. This labeled-input-to-known-output pattern is the defining characteristic of supervised classification.

Answer

Semi-supervised learning

Answer

Reinforcement learning

Answer

Unsupervised learning

Question 7

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?

Accepted Answer

Investigate the cause and retrain the model to reduce bias. 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.

Answer

Lower the decision threshold for women to equalize false positive rates

Answer

Proceed with deployment because the overall accuracy is acceptable

Answer

Use the model but require manual review for all female candidates

Question 8

A company wants to use Einstein to forecast sales beyond simple manager rollups, comparing AI predictions with rep commitments. Which feature should they enable?

Accepted Answer

Einstein Forecasting. Einstein Forecasting is the correct feature because it is specifically designed to generate AI-driven sales forecasts by analyzing historical data, pipeline trends, and external factors, then comparing those predictions against rep commitments. Unlike simple manager rollups, it provides a statistical baseline that helps identify gaps between AI predictions and human estimates, enabling more accurate revenue planning.

Answer

Einstein Discovery

Answer

Einstein Prediction Builder

Answer

Einstein Opportunity Scoring

Question 9

In Salesforce CRM Analytics (formerly Einstein Analytics), what is the primary purpose of a dataset?

Accepted Answer

To prepare data for AI and analytics. In Salesforce CRM Analytics, a dataset is the foundational data structure that transforms raw data into an optimized, columnar format for analytics and AI features like Einstein Discovery. It is created by extracting, cleaning, and aggregating data from sources such as Salesforce objects or external connectors, enabling efficient querying, dashboarding, and machine learning model training. This makes option A correct because the primary purpose is to prepare data specifically for AI and analytics workloads.

Answer

To run SQL queries directly

Answer

To store raw, unprocessed records

Answer

To create dashboards only

Question 10

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?

Accepted Answer

Retrain the model using a balanced dataset that includes equal representation from all genders and implement ongoing fairness monitoring.. The 92% overall accuracy metric masks a disparate impact on a protected group, which is a classic fairness failure in AI systems trained on biased historical data. Retraining with a balanced, representative dataset directly addresses the root cause (sampling bias in the training data), and ongoing fairness monitoring ensures the model does not drift back into biased behavior after deployment. This aligns with the Salesforce AI Associate principle of building trustworthy, ethical AI that is fair, inclusive, and continuously evaluated.

Answer

Continue using the model because 92% accuracy is acceptable and the bias is not significant.

Answer

Replace the current AI tool with a different vendor's tool without further analysis.

Answer

Manually adjust the scoring algorithm to give preference to female candidates to balance the outcome.

Question 11

A generative AI chatbot sometimes produces factually incorrect responses about a company's products. What is this phenomenon called?

Accepted Answer

Hallucination. Hallucination is the correct term for a generative AI model producing fluent but factually incorrect or fabricated output, such as inventing product features or specifications that do not exist. It occurs because LLMs generate statistically plausible token sequences rather than retrieving verified facts, so they can confidently state falsehoods. This is a well-known limitation of generative models and a key concern for enterprise deployments.

Answer

Data leakage

Answer

Overfitting

Answer

Bias

Question 12

A data engineer is troubleshooting a predictive model that stopped updating. The data flow from Data Cloud shows 'Data Transform Failed' with error: 'Field Amount cannot be null'. What is the most likely cause?

Accepted Answer

The target field in the model requires a non-null value but source data has nulls.. The error 'Field Amount cannot be null' indicates that the target field in the predictive model is configured to require a non-null value. When the data flow attempts to write records with null Amount values into that field, the transform fails. This is a common schema constraint mismatch where the source data contains nulls that violate the target field's nullability requirement.

Answer

The data transform includes a filter that removes records with null Amount.

Answer

The source object has a validation rule.

Answer

The data flow schedule is incorrect.

Question 13

Which TWO are best practices when implementing Einstein Bots? (Choose two.)

Accepted Answer

Start with high-volume, low-complexity conversations. Option A is correct because starting with high-volume, low-complexity conversations lets you deliver quick wins, keep the bot's intent model manageable, and build user trust before expanding scope. Option D is correct because Einstein Bots provides conversation logs and analytics that should be reviewed regularly to refine utterances, add training phrases, and improve intent matching accuracy over time. Option B is incorrect because complex intents increase misclassification risk and make initial training and testing harder, so they should be deferred until the bot matures. Option C is incorrect because a bot should not be expected to handle all conversations; escalation to a human agent is a required fallback for unsupported or sensitive requests. Option E is incorrect because deploying to all channels immediately without testing skips validation of intents, channel-specific behavior, and fallback handling, risking poor customer experiences.

Answer

Use complex intents for initial setup

Answer

Configure the bot to handle all customer conversations

Answer

Deploy the bot to all channels immediately without testing

Question 14

A data scientist is building a custom AI model using Salesforce Data Cloud. They want to follow best practices for data minimisation and consent management. Which THREE steps should they take?

Accepted Answer

Obtain consent from customers before using their data in the model. Option A is correct because consent management is a core Data Cloud best practice: before using customer data to train or score an AI model, the data scientist must ensure the individual's consent (e.g., via Consent Management objects and Data Cloud consent policies) permits that specific use, otherwise the processing violates data minimisation and privacy obligations. Option C is correct because documenting the data sources, Data Streams, Data Model Objects (DMOs), and calculated insights/features used in the model provides the auditability and lineage required for governance, regulatory review, and reproducibility of the model. Option D is correct because data minimisation means selecting only the fields that are relevant to the prediction task rather than ingesting or exposing unnecessary attributes, reducing privacy risk and improving model quality. Option B is not correct because including all available fields contradicts data minimisation and can introduce noise, overfitting, and unnecessary personal data processing. Option E is not correct because using synthetic data is not a required best-practice step here and does not by itself satisfy consent or minimisation requirements for real customer data.

Answer

Include all available fields in the model to maximize accuracy

Answer

Use synthetic data instead of real customer data

Question 15

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

Accepted Answer

Whether features are correlated with protected attributes.. Option B is correct because bias often enters a model when input features are correlated with protected attributes such as race, gender, or age, which can cause the model to indirectly discriminate even if those attributes are excluded. Option C is correct because evaluating performance metrics like accuracy, precision, recall, or false positive rate across demographic groups reveals disparate impact and unequal error rates that signal bias. Option E is correct because if the training data underrepresents or misrepresents certain groups, the model will learn skewed patterns and generalize poorly to those populations. Option A is not a direct bias factor, since model complexity affects capacity and overfitting but not whether outcomes are systematically unfair to a group. Option D is irrelevant to bias evaluation, as deployment recency says nothing about the fairness or representativeness of the model's data and predictions.

Answer

The complexity of the model architecture.

Answer

The date the model was last deployed.