AI Associate Hard Questions — 25 Difficult Questions

Question 1

A company is building a text classification model for customer support tickets. They have a dataset of 10,000 tickets. The team decides to use active learning for labeling. Which approach best aligns with active learning principles?

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

Accepted Answer

Train a preliminary model and prioritize labeling tickets with low prediction confidence.. Active learning iteratively selects the most informative unlabeled data points for labeling, typically those with low prediction confidence from a preliminary model. This minimizes labeling effort while maximizing model performance, which is the core principle of active learning.

Answer

Randomly select 2,000 tickets and label them manually.

Answer

Use a pre-trained model to label all tickets automatically.

Answer

Have subject matter experts label all 10,000 tickets.

Question 2

A company has international customers and wants Einstein Prediction Builder to forecast deal closure probability. The data includes fields like 'region', 'product line', and 'deal amount'. What is a best practice to ensure the model works for all regions?

Accepted Answer

Group regions into broader categories like 'Americas', 'EMEA', 'APAC'.. Grouping regions into broader categories like 'Americas', 'EMEA', and 'APAC' reduces high cardinality and sparsity in categorical features, which improves model stability and prevents overfitting in Einstein Prediction Builder. This approach ensures each region group has sufficient training data to learn meaningful patterns, enabling the model to generalize better across all regions without introducing bias from rare categories.

Answer

One-hot encode the region field using 50+ dummy variables.

Answer

Remove the region field to avoid bias.

Answer

Use region as a numeric rank based on past conversion rates.

Question 3

Universal Containers (UC) uses Einstein Lead Scoring to prioritize leads. They have 500,000 leads in the system. Recently, the model scores have been inconsistent: some leads with low activity receive high scores, while active leads score low. The model was trained 3 months ago. UC updates lead records daily via an external system, but the data is often incomplete (e.g., missing company size). Support has reported slow performance on lead views. The admin notices that the 'Data Refresh Status' for Einstein Lead Scoring shows 'Pending' for 2 weeks. UC wants to improve model accuracy and performance.

Which action should the admin take first?

Accepted Answer

Schedule a data transform to cleanse and refresh lead data before model training. The 'Data Refresh Status' has been 'Pending' for two weeks, indicating that the model is not receiving updated lead data. Scheduling a data transform to cleanse and refresh lead data before model training directly addresses the root cause: incomplete and stale data (e.g., missing company size) leads to inconsistent scores. This action ensures the model trains on clean, current data, improving both accuracy and performance.

Answer

Increase data retention period for lead records

Answer

Rebuild the Einstein Lead Scoring model using different fields

Answer

Increase the model training frequency to weekly

Question 4

An admin is configuring Einstein Lead Scoring. They want to ensure the lead score is visible in list views and reports. Which TWO settings or actions are required?

Accepted Answer

Add the Lead Score field to the list view layout. The Lead Score field must be added to the list view layout to make it visible in list views and reports. Without adding the field to the layout, users cannot see the score in those contexts, even if scoring is enabled.

Answer

Run a lead scoring batch job manually

Answer

Create a custom report type for Lead Score

Answer

Assign the 'View Lead Score' permission to users

Question 5

A service team uses Einstein Discovery to analyze customer churn. The story shows 'Average Resolution Time' is a key driver. What is the best action?

Accepted Answer

Reduce Average Resolution Time through process changes. Einstein Discovery identifies drivers; reducing resolution time directly addresses root cause. Custom reports just show data, flows send alerts, processes update records but don't reduce time.

Answer

Update account records via process builder

Answer

Configure a flow to send churn alerts

Answer

Create a custom report on churn

Question 6

Which THREE capabilities are provided by Einstein GPT in Sales and Service?

Accepted Answer

Summarizing case conversations from service interactions.. Einstein GPT leverages generative AI to automatically summarize case conversations from service interactions, enabling agents to quickly grasp key details without reading entire transcripts. This capability is built on Salesforce's proprietary AI models that process natural language from service records to produce concise, actionable summaries.

Answer

Generating report charts and dashboards.

Answer

Creating workflow rules based on user behavior.

Question 7

A developer is building an Einstein Bot that needs to understand when a customer says 'I want to return a purchase' and route them to the returns process. How should they configure the bot?

Accepted Answer

Create an intent called 'Return Purchase' and train it with sample phrases. In Einstein Bots, intents represent the customer's goal, and entities capture specifics. 'Return purchase' is an intent, not an entity. The bot uses NLP to match utterances to intents.

Answer

Create a dialogue that triggers on the exact phrase 'I want to return a purchase'

Answer

Create an entity called 'Return Purchase' and map it to a dialogue

Answer

Use an intent called 'Customer Service' and a custom entity for return

Question 8

An admin is configuring Einstein Prediction Builder to predict case escalation. Which TWO components must be selected during setup?

Accepted Answer

Prediction field (binary classification target). Einstein Prediction Builder requires a binary classification target field to define the outcome being predicted—in this case, whether a case will escalate. This field must have exactly two distinct values (e.g., 'Yes'/'No' or 0/1) to train the model. Without specifying the prediction field, the builder cannot determine what event to forecast.

Answer

Prediction explanation template

Answer

Data set (records to train on)

Answer

Prediction score field name

Question 9

A financial services company uses Salesforce AI to detect fraudulent transactions. The dataset has 1 million legitimate transactions and only 1,000 fraudulent ones. The model trained with default parameters achieves 99.9% accuracy but identifies no fraud (precision and recall of 0). The data scientist wants to maximize fraud detection (recall) while minimizing false positives. Which approach is most effective?

Accepted Answer

Use SMOTE to generate synthetic fraud samples to balance the dataset.. With extreme imbalance, oversampling the minority class (e.g., SMOTE) generates synthetic fraud examples, helping the model learn fraud patterns and improve recall without discarding legitimate data.

Answer

Increase the weight of the majority class in the loss function.

Answer

Train multiple models on different random subsets and average predictions.

Answer

Use a simpler model to avoid overfitting on the majority class.

Question 10

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

Accepted Answer

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

Answer

Accuracy

Answer

Privacy

Answer

Transparency

Question 11

A financial services company is deploying Einstein AI and must comply with regulations requiring explainable decisions. Which Einstein capability allows them to understand why an AI model made a specific prediction?

Accepted Answer

Einstein Trust Layer with model explainability features.. The Einstein Trust Layer includes model explainability features that provide insights into why a specific prediction was made, such as highlighting the key input features that influenced the outcome. This directly addresses regulatory requirements for explainable AI decisions by offering transparency into the model's reasoning process.

Answer

Data Mask to protect sensitive information.

Answer

Salesforce Shield with encryption and monitoring.

Answer

Field Audit Trail to track changes to data.

Question 12

A data scientist is using Einstein Vision and Language Platform for text classification. They need to handle custom entities (NER) and classify text into multiple categories. Which THREE capabilities of the Einstein Platform Services API should they use?

Accepted Answer

Sentiment Analysis. Sentiment Analysis is a key capability of the Einstein Platform Services API that allows the data scientist to determine the emotional tone (positive, negative, or neutral) of text, which is essential for understanding customer feedback or social media posts. This complements the other required capabilities—Text Classification for categorizing text into multiple categories and Named Entity Recognition (NER) for extracting custom entities—forming a complete solution for the described text classification and NER tasks.

Answer

Object Detection

Answer

Image Classification

Question 13

A company uses an NLP model to detect intent in customer messages. The model works well for English but fails for Spanish messages. What is the most likely cause?

Accepted Answer

The training data is predominantly English. If the training data lacks Spanish examples, the model cannot learn Spanish patterns.

Answer

The model is overfitted to English

Answer

The inference threshold is too high

Answer

The model uses the wrong algorithm

Question 14

A company wants to build an autonomous AI agent in Salesforce that can handle customer returns, refunds, and exchanges without human intervention. Which THREE components are required to build this agent using Agentforce?

Accepted Answer

Topics and Actions. To build an autonomous AI agent in Salesforce that handles customer returns, refunds, and exchanges without human intervention, you need Topics and Actions to define the specific business processes (e.g., 'Process Return') and the corresponding API calls or flows, Agent Builder to configure the agent's behavior and link it to those topics, and Testing in Agent Builder to validate the agent's responses and ensure it operates correctly before deployment.

Answer

Prompt Builder

Answer

Einstein Copilot

Question 15

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

Accepted Answer

Use majority voting among multiple labelers.. 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.

Answer

Use a single expert labeler for all data.

Answer

Label all data by a single expert labeler.

Answer

Allow each labeler to interpret guidelines freely.

Question 16

According to Salesforce's AI ethics principles, which three pillars should guide the development of AI applications?

Accepted Answer

Accountability.. Accountability is one of Salesforce's core AI ethics principles, requiring that organizations take responsibility for the outcomes of their AI systems. This principle ensures that there is human oversight and that AI applications are designed with mechanisms for redress and governance, aligning with Salesforce's commitment to ethical AI development.

Answer

Profitability.

Answer

Scalability.

Question 17

A service organization wants to deploy an Einstein Bot to handle common support inquiries. They need to define the bot's conversational flow and train it to understand user requests. Which THREE components must be configured in the bot builder?

Accepted Answer

Intents. Intents (A) are correct because they define the purpose or goal of a user's input, such as 'Check Order Status' or 'Reset Password'. In the Einstein Bot Builder, intents map user utterances to specific bot actions, enabling the bot to understand and route requests appropriately. Without intents, the bot cannot classify what the user wants.

Answer

Prediction scores

Answer

Training phrases

Question 18

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

Accepted Answer

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

Answer

Use the AI model as the sole decision-maker.

Answer

Use all available data including protected attributes.

Answer

Ignore adverse impact if the model is accurate.

Question 19

Refer to the exhibit. A developer wrote a trigger to call an Einstein prediction API on lead insert. When new leads are created, the trigger fails with a 'Too many SOQL queries' error. What is the most likely cause?

Accepted Answer

The trigger is making API calls in a loop, causing governor limit issues. The trigger is making synchronous API calls to the Einstein prediction service inside a loop that iterates over each new lead. Each API call counts against the 'Number of callouts' governor limit (default 100 per transaction), and if more leads are inserted than the limit allows, the transaction fails with a 'Too many SOQL queries' error—though the error message is misleading, the root cause is exceeding the callout limit, not SOQL queries. The trigger should batch the API calls or use asynchronous processing to stay within limits.

Answer

The trigger is executed before the lead is saved

Answer

The trigger is querying a related object without an index

Answer

The Score__c field is not writeable

Question 20

Which THREE are key dimensions of data quality that directly impact AI model performance?

Accepted Answer

Consistency.. Consistency is a key dimension of data quality because AI models rely on stable patterns in the data. If the same entity is represented differently across records (e.g., 'NY' vs 'New York'), the model may learn incorrect correlations, leading to degraded prediction accuracy and unreliable outputs.

Answer

Timeliness.

Answer

Data volume.