SALESFORCE AI ASSOCIATE (SF-AI) CERTIFICATION COMPLETE
EXAM READY | VERIFIED QUESTIONS AND ANSWERS |
COMPREHENSIVE LATEST VERSION 2026/2027
SF-AI: SALESFORCE AI ASSOCIATE OFFICIAL
Q1. What is the primary goal of supervised learning? A. To find hidden
patterns in unlabeled data
B. To learn a mapping from inputs to outputs using labeled training data
C. To maximize rewards through trial and error
D. To reduce dimensionality in datasets
ANSWER : B
Q2. Which Salesforce AI feature uses natural language processing to
analyze text sentiment? A. Einstein Prediction Builder
B. Einstein Sentiment Analysis
C. Einstein Vision
D. Einstein Discovery
ANSWER : B
Q3. What does "training data" refer to in machine learning? A. Data
used to test the final model performance
B. Data used to teach the model patterns during development
C. Data collected from production environments
D. Data that has been encrypted for security
ANSWER : B
Q4. Which type of AI model is best suited for predicting a continuous
numerical value? A. Classification model
B. Regression model
C. Clustering model
D. Reinforcement learning model
ANSWER : B
Q5. What is "overfitting" in machine learning? A. When a model
performs well on training data but poorly on new data
,B. When a model is too simple to capture data patterns
C. When training takes too long
D. When a model uses too little data
ANSWER : A
Q6. What is the purpose of a validation dataset? A. To train the model
initially
B. To tune hyperparameters and prevent overfitting
C. To deploy the model to production
D. To store backup copies of training data
ANSWER : B
Q7. Which Salesforce Einstein feature provides product
recommendations? A. Einstein Next Best Action
B. Einstein Recommendation Builder
C. Einstein Prediction Builder
D. Einstein Bots
ANSWER : B
Q8. What is "feature engineering"? A. Creating new input variables from
raw data to improve model performance
B. Building the user interface for AI applications
C. Writing code for neural networks
D. Managing data storage infrastructure
ANSWER : A
Q9. What is the difference between AI and machine learning? A. AI is
a subset of machine learning
B. Machine learning is a subset of AI
C. They are completely unrelated technologies
D. AI only works with structured data
ANSWER : B
Q10. Which algorithm is commonly used for classification tasks? A.
Linear regression
B. Decision trees
C. K-means clustering
D. Principal component analysis
ANSWER : B
Q11. What does "inference" mean in the context of AI models? A.
Training a model on historical data
,B. Using a trained model to make predictions on new data
C. Collecting data from multiple sources
D. Validating data quality
ANSWER : B
Q12. What is a "neural network"? A. A network of computers sharing
processing power
B. A machine learning model inspired by biological neurons
C. A social network for AI researchers
D. A database of neural medical data
ANSWER : B
Q13. Which Salesforce tool allows admins to build custom AI
predictions without code? A. Apex
B. Einstein Prediction Builder
C. Lightning Web Components
D. Data Loader
ANSWER : B
Q14. What is "natural language processing" (NLP)? A. Processing
images to understand content
B. Enabling computers to understand and generate human language
C. Processing numerical data for statistical analysis
D. Managing network protocols
ANSWER : B
Q15. What is the purpose of regularization in machine learning? A.
To increase model complexity
B. To prevent overfitting by penalizing complex models
C. To speed up training time
D. To increase the size of the dataset
ANSWER : B
Q16. Which type of learning involves an agent learning through
rewards and penalties? A. Supervised learning
B. Unsupervised learning
C. Reinforcement learning
D. Semi-supervised learning
ANSWER : C
Q17. What is a "confusion matrix" used for? A. Encrypting sensitive
data
, B. Evaluating classification model performance
C. Storing training datasets
D. Managing user permissions
ANSWER : B
Q18. What does "precision" measure in a classification model? A. The
proportion of true positives among all positive predictions
B. The proportion of true positives among all actual positives
C. The overall accuracy of the model
D. The speed of model training
ANSWER : A
Q19. What is "recall" in machine learning evaluation? A. The
proportion of true positives identified out of all actual positives
B. The proportion of correct predictions overall
C. The time taken to retrieve data
D. The memory used by the model
ANSWER : A
Q20. Which Salesforce AI capability can classify images? A. Einstein
Language
B. Einstein Vision
C. Einstein Analytics
D. Einstein Voice
ANSWER : B
Q21. What is "deep learning"? A. Learning that takes a long time
B. Machine learning using neural networks with multiple layers
C. Learning from very small datasets
D. Manual data entry processes
ANSWER : B
Q22. What is the purpose of cross-validation? A. To deploy models
faster
B. To assess model performance across different data subsets
C. To encrypt data during transmission
D. To create backup copies of models
ANSWER : B
Q23. What is a "hyperparameter"? A. A parameter learned during
model training
B. A configuration set before training begins
EXAM READY | VERIFIED QUESTIONS AND ANSWERS |
COMPREHENSIVE LATEST VERSION 2026/2027
SF-AI: SALESFORCE AI ASSOCIATE OFFICIAL
Q1. What is the primary goal of supervised learning? A. To find hidden
patterns in unlabeled data
B. To learn a mapping from inputs to outputs using labeled training data
C. To maximize rewards through trial and error
D. To reduce dimensionality in datasets
ANSWER : B
Q2. Which Salesforce AI feature uses natural language processing to
analyze text sentiment? A. Einstein Prediction Builder
B. Einstein Sentiment Analysis
C. Einstein Vision
D. Einstein Discovery
ANSWER : B
Q3. What does "training data" refer to in machine learning? A. Data
used to test the final model performance
B. Data used to teach the model patterns during development
C. Data collected from production environments
D. Data that has been encrypted for security
ANSWER : B
Q4. Which type of AI model is best suited for predicting a continuous
numerical value? A. Classification model
B. Regression model
C. Clustering model
D. Reinforcement learning model
ANSWER : B
Q5. What is "overfitting" in machine learning? A. When a model
performs well on training data but poorly on new data
,B. When a model is too simple to capture data patterns
C. When training takes too long
D. When a model uses too little data
ANSWER : A
Q6. What is the purpose of a validation dataset? A. To train the model
initially
B. To tune hyperparameters and prevent overfitting
C. To deploy the model to production
D. To store backup copies of training data
ANSWER : B
Q7. Which Salesforce Einstein feature provides product
recommendations? A. Einstein Next Best Action
B. Einstein Recommendation Builder
C. Einstein Prediction Builder
D. Einstein Bots
ANSWER : B
Q8. What is "feature engineering"? A. Creating new input variables from
raw data to improve model performance
B. Building the user interface for AI applications
C. Writing code for neural networks
D. Managing data storage infrastructure
ANSWER : A
Q9. What is the difference between AI and machine learning? A. AI is
a subset of machine learning
B. Machine learning is a subset of AI
C. They are completely unrelated technologies
D. AI only works with structured data
ANSWER : B
Q10. Which algorithm is commonly used for classification tasks? A.
Linear regression
B. Decision trees
C. K-means clustering
D. Principal component analysis
ANSWER : B
Q11. What does "inference" mean in the context of AI models? A.
Training a model on historical data
,B. Using a trained model to make predictions on new data
C. Collecting data from multiple sources
D. Validating data quality
ANSWER : B
Q12. What is a "neural network"? A. A network of computers sharing
processing power
B. A machine learning model inspired by biological neurons
C. A social network for AI researchers
D. A database of neural medical data
ANSWER : B
Q13. Which Salesforce tool allows admins to build custom AI
predictions without code? A. Apex
B. Einstein Prediction Builder
C. Lightning Web Components
D. Data Loader
ANSWER : B
Q14. What is "natural language processing" (NLP)? A. Processing
images to understand content
B. Enabling computers to understand and generate human language
C. Processing numerical data for statistical analysis
D. Managing network protocols
ANSWER : B
Q15. What is the purpose of regularization in machine learning? A.
To increase model complexity
B. To prevent overfitting by penalizing complex models
C. To speed up training time
D. To increase the size of the dataset
ANSWER : B
Q16. Which type of learning involves an agent learning through
rewards and penalties? A. Supervised learning
B. Unsupervised learning
C. Reinforcement learning
D. Semi-supervised learning
ANSWER : C
Q17. What is a "confusion matrix" used for? A. Encrypting sensitive
data
, B. Evaluating classification model performance
C. Storing training datasets
D. Managing user permissions
ANSWER : B
Q18. What does "precision" measure in a classification model? A. The
proportion of true positives among all positive predictions
B. The proportion of true positives among all actual positives
C. The overall accuracy of the model
D. The speed of model training
ANSWER : A
Q19. What is "recall" in machine learning evaluation? A. The
proportion of true positives identified out of all actual positives
B. The proportion of correct predictions overall
C. The time taken to retrieve data
D. The memory used by the model
ANSWER : A
Q20. Which Salesforce AI capability can classify images? A. Einstein
Language
B. Einstein Vision
C. Einstein Analytics
D. Einstein Voice
ANSWER : B
Q21. What is "deep learning"? A. Learning that takes a long time
B. Machine learning using neural networks with multiple layers
C. Learning from very small datasets
D. Manual data entry processes
ANSWER : B
Q22. What is the purpose of cross-validation? A. To deploy models
faster
B. To assess model performance across different data subsets
C. To encrypt data during transmission
D. To create backup copies of models
ANSWER : B
Q23. What is a "hyperparameter"? A. A parameter learned during
model training
B. A configuration set before training begins
