Section 1: Introduction to Artificial Intelligence (50

MCQs)
1. AI stands for:
a) Automatic Intelligence
b) Artificial Intelligence
c) Advanced Intelligence
d) Algorithmic Intelligence
Answer: b
2. AI is mainly focused on:
a) Creating human-like intelligence in machines
b) Building faster processors
c) Reducing energy consumption
d) Storing large datasets
Answer: a
3. The term "Artificial Intelligence" was coined by:
a) Alan Turing
b) John McCarthy
c) Marvin Minsky
d) Geoffrey Hinton
Answer: b
4. AI is used in:
a) Robotics
b) Natural Language Processing
c) Image Recognition
d) All of the above
Answer: d
5. Strong AI is also called:
a) General AI
b) Narrow AI
c) Weak AI
d) Applied AI
Answer: a
6. Weak AI is also called:
a) Narrow AI
b) General AI
c) Cognitive AI
d) Hybrid AI
Answer: a
7. Which of the following is an example of Weak AI?
a) ChatGPT
b) Self-aware AI
c) Human-level intelligence machine
d) None
Answer: a
8. Strong AI can:
a) Perform only specific tasks
b) Understand, learn, and reason like humans
c) Only calculate numbers
 d) Only recognize images
Answer: b
9. AI systems can be classified as:
a) Symbolic AI and Sub-symbolic AI
b) Weak AI and Strong AI
c) Both a and b
d) None
Answer: c
10. The Turing Test checks:
a) Machine speed
b) Intelligence level of a machine
c) Energy efficiency
d) Storage capacity
Answer: b
11. AI applications include:
a) Healthcare
b) Finance
c) Autonomous vehicles
d) All of the above
Answer: d
12. AI that simulates human reasoning using rules is called:
a) ML-based AI
b) Rule-based AI
c) Neural AI
d) Strong AI
Answer: b
13. Rule-based AI uses:
a) Data only
b) Hard-coded rules
c) Neural networks
d) Genetic algorithms
Answer: b
14. Rule-based AI is limited because:
a) It cannot learn from data
b) It is too fast
c) It requires large storage
d) It is always weak AI
Answer: a
15. Rule-based AI works best when:
a) Problems are simple and structured
b) Problems are complex
c) Data is dynamic
d) None
Answer: a
16. Expert systems are examples of:
a) ML-based AI
b) Rule-based AI
c) Neural networks
d) Robotics
Answer: b
17. ML-based AI differs from Rule-based AI because:
a) It uses fixed rules
b) It learns from data
c) It cannot adapt
d) It is weak AI only
Answer: b
18. ML stands for:
a) Machine Logic
b) Machine Learning
c) Modeling Learning
d) Multi-layered AI
Answer: b
19. Machine learning allows AI to:
a) Learn from data and improve over time
b) Follow only hard-coded instructions
c) Solve only arithmetic problems
d) Be manually programmed for every task
Answer: a
20. Supervised learning requires:
a) Labeled data
b) Unlabeled data
c) No data
d) Only test data
Answer: a
21. Unsupervised learning uses:
a) Labeled data
b) Unlabeled data
c) Reinforcement signals
d) Rules only
Answer: b
22. Classification is a type of:
a) Supervised learning
b) Unsupervised learning
c) Reinforcement learning
d) Rule-based AI
Answer: a
23. Regression is used to:
a) Predict continuous values
b) Predict categories
c) Cluster data
d) Encode rules
Answer: a
24. Clustering is a type of:
a) Supervised learning
b) Unsupervised learning
c) Reinforcement learning
d) Rule-based AI
Answer: b
25. An example of supervised learning is:
a) Spam email detection
 b) Customer segmentation
c) Market basket analysis
d) None
Answer: a
26. An example of unsupervised learning is:
a) Handwritten digit recognition
b) Clustering customers by purchasing behavior
c) Predicting house prices
d) Sentiment analysis
Answer: b
27. Reinforcement learning involves:
a) Learning from rewards and penalties
b) Learning from labeled data
c) Learning from clustering
d) Learning only from rules
Answer: a
28. Neural networks are inspired by:
a) Human brain
b) Classical programming
c) Rule-based logic
d) Probability theory
Answer: a
29. Deep learning is a subset of:
a) Artificial Intelligence
b) Machine Learning
c) Neural Networks
d) All of the above
Answer: b
30. Feature engineering is:
a) Selecting important input variables
b) Creating rules
c) Reinforcement signals
d) Neural processing
Answer: a
31. Training dataset is used for:
a) Model training
b) Model testing
c) Model deployment
d) Data cleaning
Answer: a
32. Testing dataset is used for:
a) Training
b) Evaluating model performance
c) Feature selection
d) Preprocessing
Answer: b
33. Overfitting occurs when:
a) Model performs well on training but poorly on test data
b) Model underperforms on training data
c) Model uses fewer features
 d) Model is simple
Answer: a
34. Underfitting occurs when:
a) Model cannot capture underlying trend
b) Model memorizes data
c) Model uses too many features
d) Model is perfect
Answer: a
35. Accuracy measures:
a) Correct predictions / total predictions
b) False positives only
c) False negatives only
d) Training speed
Answer: a
36. Precision measures:
a) Correct positive predictions / total predicted positives
b) Correct negative predictions / total predicted negatives
c) Total predictions / correct predictions
d) Model complexity
Answer: a
37. Recall measures:
a) Correct positive predictions / total actual positives
b) Correct negative predictions / total actual negatives
c) Predicted positives / actual positives
d) Training error
Answer: a
38. F1-score is:
a) Harmonic mean of precision and recall
b) Arithmetic mean of precision and recall
c) Difference of precision and recall
d) Ratio of precision to recall
Answer: a
39. AI in healthcare can be used for:
a) Disease prediction
b) Image analysis
c) Personalized treatment
d) All of the above
Answer: d
40. AI in finance can be used for:
a) Fraud detection
b) Algorithmic trading
c) Risk assessment
d) All of the above
Answer: d
41. Autonomous vehicles use AI for:
a) Navigation
b) Obstacle detection
c) Traffic prediction
d) All of the above
Answer: d
42. Natural Language Processing (NLP) is:
a) Understanding human language
b) Recognizing images
c) Controlling robots
d) Clustering data
Answer: a
43. Chatbots use:
a) NLP
b) ML
c) Both a and b
d) None
Answer: c
44. Recommendation systems use:
a) Supervised learning
b) Unsupervised learning
c) Both
d) None
Answer: c
45. AI in agriculture can help with:
a) Crop disease detection
b) Yield prediction
c) Soil analysis
d) All of the above
Answer: d
46. AI in manufacturing helps with:
a) Predictive maintenance
b) Quality control
c) Process optimization
d) All of the above
Answer: d
47. Expert systems are widely used in:
a) Medicine
b) Law
c) Engineering
d) All of the above
Answer: d
48. Self-learning AI can:
a) Learn from experience
b) Follow fixed rules
c) Only classify images
d) Only predict numbers
Answer: a
49. AI ethics deals with:
a) Fairness, transparency, and accountability
b) Model accuracy
c) Training speed
d) Data storage
Answer: a
50. AI challenges include:
a) Bias in data
 b) High computation cost
c) Interpretability
d) All of the above
Answer: d

Section 2: Strong AI and Weak AI (51–100 MCQs)

51. Strong AI aims to:
a) Perform narrow, specific tasks
b) Mimic human intelligence fully
c) Only calculate numbers
d) Recognize images only
Answer: b
52. Weak AI is also known as:
a) General AI
b) Narrow AI
c) Self-aware AI
d) Cognitive AI
Answer: b
53. Strong AI is:
a) Already achieved
b) A future goal
c) Same as Weak AI
d) Not related to intelligence
Answer: b
54. Weak AI can:
a) Reason like humans
b) Perform tasks intelligently but narrowly
c) Become conscious
d) Solve all problems
Answer: b
55. Siri and Alexa are examples of:
a) Strong AI
b) Weak AI
c) Artificial Consciousness
d) General Intelligence
Answer: b
56. Strong AI would require:
a) Advanced algorithms only
b) Self-awareness and reasoning
c) Data storage
d) Basic programming
Answer: b
57. Weak AI focuses on:
a) Specific problem-solving
b) Human-like general intelligence
c) Self-learning
d) Consciousness
Answer: a
58. Strong AI is also called:
a) Cognitive AI
b) Artificial General Intelligence (AGI)
c) Rule-based AI
d) Narrow AI
Answer: b
59. The main challenge of Strong AI is:
a) Limited data
b) Machine consciousness
c) Slow computers
d) Lack of programming languages
Answer: b
60. Weak AI can beat humans at:
a) Narrow tasks like chess or image recognition
b) All tasks
c) Emotional reasoning
d) Self-awareness
Answer: a
61. AGI aims to:
a) Solve a single task
b) Perform any intellectual task like humans
c) Only recognize patterns
d) Only play games
Answer: b
62. Current AI is mostly:
a) Strong AI
b) Weak AI
c) Conscious AI
d) None
Answer: b
63. Turing Test can assess:
a) Weak AI only
b) Strong AI capability
c) ML models
d) Robotics hardware
Answer: b
64. Examples of Weak AI include:
a) Chess-playing programs
b) Autonomous cars
c) Recommendation systems
d) All of the above
Answer: d
65. Strong AI would require:
a) Human-level reasoning
b) Labeled datasets
c) Rule-based programming only
d) Supervised ML
Answer: a
66. Weak AI limitation:
a) Cannot think beyond predefined tasks
 b) Can become conscious
c) Can reason like humans
d) Can learn without data
Answer: a
67. AI that understands and interprets context is:
a) Weak AI
b) Strong AI
c) Rule-based AI
d) ML-based AI
Answer: b
68. Weak AI uses:
a) Hard-coded rules
b) Machine learning models
c) Both a and b
d) Self-awareness
Answer: c
69. Strong AI is also referred to as:
a) Narrow AI
b) General AI
c) Expert system
d) Symbolic AI
Answer: b
70. Weak AI focuses on:
a) Specific applications
b) Human-like intelligence
c) Consciousness
d) Creativity
Answer: a
71. AGI is a type of:
a) Weak AI
b) Strong AI
c) Rule-based AI
d) Expert system
Answer: b
72. Current self-driving cars use:
a) Weak AI
b) Strong AI
c) Human-level AI
d) Conscious AI
Answer: a
73. Weak AI can be:
a) Task-specific
b) Adaptive within task
c) Limited to training data
d) All of the above
Answer: d
74. Strong AI is hypothesized to:
a) Understand and reason across domains
b) Only calculate data
c) Only perform image recognition
 d) Only follow rules
Answer: a
75. Example of AGI in fiction:
a) HAL 9000
b) Siri
c) Alexa
d) Google Maps
Answer: a
76. Narrow AI cannot:
a) Beat humans at games
b) Understand general knowledge
c) Recognize patterns
d) Predict outcomes in a domain
Answer: b
77. Weak AI systems are:
a) Ubiquitous today
b) Rare and experimental
c) Self-conscious
d) AGI
Answer: a
78. Strong AI would likely require:
a) Advanced computation and algorithms
b) Human-like reasoning capabilities
c) Ability to learn and adapt across tasks
d) All of the above
Answer: d
79. Weak AI successes are mostly in:
a) Healthcare, finance, and robotics
b) AGI research
c) Self-aware AI development
d) General reasoning
Answer: a
80. Weak AI cannot:
a) Exceed humans in specific tasks
b) Generalize across all tasks
c) Learn from data
d) Predict patterns
Answer: b
81. AI that cannot learn or adapt is:
a) Rule-based AI
b) ML-based AI
c) Strong AI
d) Reinforcement AI
Answer: a
82. Rule-based AI can be improved by:
a) Adding more rules
b) Machine learning
c) Both a and b
d) None
Answer: c
83. Expert systems are:
a) Weak AI
b) Strong AI
c) ML-based AI
d) Conscious AI
Answer: a
84. AI systems today are mostly:
a) Weak AI / Narrow AI
b) Strong AI
c) Conscious AI
d) Human-like AI
Answer: a
85. Self-learning AI still falls under:
a) Weak AI
b) Strong AI
c) AGI
d) Conscious AI
Answer: a
86. Weak AI is designed to:
a) Solve specific problems efficiently
b) Achieve human-level general intelligence
c) Develop consciousness
d) Reason across all domains
Answer: a
87. Strong AI research focuses on:
a) General reasoning
b) Task-specific learning
c) Rule addition
d) ML optimization
Answer: a
88. Weak AI performance depends on:
a) Quality of data and models
b) Consciousness
c) Creativity
d) None
Answer: a
89. AGI aims to combine:
a) Learning, reasoning, and perception
b) Only rules
c) Only ML models
d) Only robotics
Answer: a
90. Weak AI cannot:
a) Understand context across domains
b) Predict patterns in data
c) Automate specific tasks
d) Assist humans
Answer: a
91. Strong AI would require:
a) Self-awareness
 b) Reasoning
c) Learning across tasks
d) All of the above
Answer: d
92. Weak AI examples:
a) AlphaGo
b) Grammarly
c) Google Translate
d) All of the above
Answer: d
93. Strong AI is still:
a) Theoretical
b) Widely deployed
c) Replaced Weak AI
d) Available commercially
Answer: a
94. Weak AI limitations include:
a) Cannot generalize
b) Cannot achieve consciousness
c) Both a and b
d) None
Answer: c
95. AGI is expected to:
a) Think, reason, and solve any problem like humans
b) Only perform chess
c) Only classify images
d) Follow fixed rules
Answer: a
96. Weak AI success depends on:
a) Algorithm and data
b) Self-awareness
c) Creativity
d) Consciousness
Answer: a
97. Rule-based AI is mostly:
a) Weak AI
b) Strong AI
c) AGI
d) Deep learning
Answer: a
98. Weak AI can improve over time with:
a) More data
b) Better algorithms
c) Feature engineering
d) All of the above
Answer: d
99. AI that mimics human thought is:
a) Strong AI
b) Weak AI
c) Rule-based AI
 d) ML-based AI
Answer: a
100. Most AI systems in use today are:
a) Weak AI / Narrow AI
b) Strong AI
c) Self-aware AI
d) Conscious AI
Answer: a

Section 3: Rule-based AI and ML-based AI (101–150

MCQs)
101. Rule-based AI works by:
a) Learning from data
b) Following predefined rules
c) Neural processing
d) Reinforcement learning
Answer: b
102. Expert systems are an example of:
a) Rule-based AI
b) ML-based AI
c) Deep learning
d) Reinforcement AI
Answer: a
103. ML-based AI differs from rule-based AI because it:
a) Uses fixed rules only
b) Learns patterns from data
c) Cannot adapt
d) Works without data
Answer: b
104. Decision trees are part of:
a) ML-based AI
b) Rule-based AI
c) Weak AI only
d) Strong AI only
Answer: a
105. Knowledge base is used in:
a) Rule-based AI
b) ML-based AI
c) Reinforcement learning
d) Neural networks
Answer: a
106. Inference engine is a component of:
a) ML-based AI
b) Rule-based AI
c) Neural networks
d) Deep learning
Answer: b
107. Rule-based AI is suitable for:
a) Structured, deterministic problems
b) Unstructured problems
c) Image recognition
d) Natural language understanding
Answer: a
108. ML-based AI can:
a) Learn from examples
b) Follow predefined rules
c) Only use expert knowledge
d) None
Answer: a
109. Rule-based AI is limited because:
a) Cannot adapt to new situations
b) Requires large datasets
c) Uses neural networks
d) Can perform unsupervised learning
Answer: a
110. ML-based AI is better than rule-based AI for:
a) Dynamic and complex problems
b) Fixed rule systems
c) Simple decision-making
d) Tabular calculations
Answer: a
111. Neural networks are an example of:
a) ML-based AI
b) Rule-based AI
c) Symbolic AI
d) Expert systems
Answer: a
112. ML-based AI requires:
a) Training data
b) Rules only
c) Logic programming
d) None
Answer: a
113. Reinforcement learning is part of:
a) ML-based AI
b) Rule-based AI
c) Expert systems
d) Knowledge-based AI
Answer: a
114. Supervised learning is a type of:
a) ML-based AI
b) Rule-based AI
c) Expert system
d) Symbolic AI
Answer: a
115. Rule-based AI performs well when:
a) Rules are complete and clear
 b) Data is dynamic
c) Environment changes frequently
d) Tasks are unstructured
Answer: a
116. ML-based AI can improve:
a) Automatically with more data
b) Only with rule updates
c) Only manually
d) Cannot improve
Answer: a
117. Expert systems use:
a) If-then rules
b) Neural networks
c) Clustering
d) Regression
Answer: a
118. ML-based AI can handle:
a) Uncertainty and noise
b) Only deterministic tasks
c) Fixed logic
d) None
Answer: a
119. Rule-based AI does not:
a) Learn from experience
b) Apply logic
c) Make decisions
d) Use expert knowledge
Answer: a
120. ML-based AI learns:
a) Patterns, trends, and features from data
b) Fixed rules
c) Only symbolic knowledge
d) None
Answer: a
121. Classification problems use:
a) ML-based AI
b) Rule-based AI
c) Expert systems
d) Logic-based AI
Answer: a
122. Regression problems are solved using:
a) ML-based AI
b) Rule-based AI
c) Knowledge base
d) Decision rules only
Answer: a
123. Unsupervised learning examples include:
a) Clustering
b) Decision trees
c) Expert rules
 d) Neural nets only
Answer: a
124. Rule-based AI can be combined with ML in:
a) Hybrid AI systems
b) Pure ML systems
c) Neural networks
d) Clustering only
Answer: a
125. ML-based AI can adapt to:
a) New patterns automatically
b) Predefined rules only
c) Only expert decisions
d) Fixed tasks
Answer: a
126. Fuzzy logic is an example of:
a) Rule-based AI
b) ML-based AI
c) Hybrid AI
d) Deep learning
Answer: a
127. Deep learning networks are a type of:
a) ML-based AI
b) Rule-based AI
c) Symbolic AI
d) Expert system
Answer: a
128. Rule-based AI cannot:
a) Generalize beyond its rules
b) Solve deterministic tasks
c) Use logic
d) Make decisions
Answer: a
129. ML-based AI is widely used in:
a) Image recognition
b) Speech recognition
c) Recommendation systems
d) All of the above
Answer: d
130. ML-based AI needs:
a) Feature extraction
b) Data preprocessing
c) Model selection
d) All of the above
Answer: d
131. Supervised ML requires:
a) Input-output pairs
b) Only inputs
c) Only outputs
d) Rules
Answer: a
132. Unsupervised ML does not require:
a) Labels
b) Features
c) Training data
d) Algorithms
Answer: a
133. Rule-based AI updates require:
a) Manual addition of rules
b) Automatic learning
c) Feature selection
d) Data normalization
Answer: a
134. ML-based AI updates require:
a) Retraining with new data
b) Adding rules manually
c) None
d) Only feature engineering
Answer: a
135. Hybrid AI combines:
a) Rule-based and ML-based approaches
b) Only supervised learning
c) Only unsupervised learning
d) None
Answer: a
136. Expert systems often use:
a) Inference engines
b) Neural networks
c) Clustering
d) Regression models
Answer: a
137. ML-based AI can handle:
a) Noise and incomplete data
b) Only deterministic inputs
c) Fixed logic
d) None
Answer: a
138. Rule-based AI is more explainable than ML because:
a) Rules are human-readable
b) Neural networks are transparent
c) Data is always clear
d) Algorithms are simple
Answer: a
139. ML-based AI is often called:
a) Data-driven AI
b) Rule-driven AI
c) Knowledge-driven AI
d) Logic-based AI
Answer: a
140. ML-based AI can be used for:
a) Predictive analytics
 b) Decision-making
c) Pattern recognition
d) All of the above
Answer: d
141. Knowledge representation is key for:
a) Rule-based AI
b) ML-based AI
c) Deep learning
d) Reinforcement learning
Answer: a
142. Rule-based AI may fail if:
a) Rules are incomplete
b) Environment changes
c) Tasks are dynamic
d) All of the above
Answer: d
143. Neural networks excel at:
a) Image and speech recognition
b) Logic-based rule processing
c) Simple if-then rules
d) Knowledge encoding only
Answer: a
144. ML-based AI uses algorithms like:
a) Decision trees, SVM, KNN
b) Fixed rules
c) Only inference engines
d) None
Answer: a
145. Rule-based AI is commonly used in:
a) Diagnostic systems
b) Voice assistants
c) Recommendation engines
d) Image classification
Answer: a
146. ML-based AI is commonly used in:
a) Predictive maintenance
b) Image classification
c) Fraud detection
d) All of the above
Answer: d
147. Expert systems rely on:
a) Human experts and rules
b) Data patterns
c) Neural nets
d) Deep learning
Answer: a
148. ML-based AI can generalize to:
a) New unseen data
b) Only training data
c) Fixed rules
 d) Deterministic outputs only
Answer: a
149. Rule-based AI is often:
a) Transparent and explainable
b) Black-box
c) Data-driven
d) Adaptive
Answer: a
150. ML-based AI is often:
a) Data-driven and adaptive
b) Fixed and transparent
c) Rule-driven only
d) Deterministic
Answer: a

Section 4: Machine Learning Concepts (Supervised &

Unsupervised) (151–200 MCQs)
151. Supervised learning uses:
a) Labeled data
b) Unlabeled data
c) No data
d) Only rules
Answer: a
152. Unsupervised learning uses:
a) Labeled data
b) Unlabeled data
c) Test data only
d) Rule sets
Answer: b
153. Classification is a type of:
a) Supervised learning
b) Unsupervised learning
c) Reinforcement learning
d) Rule-based AI
Answer: a
154. Regression is used to:
a) Predict continuous values
b) Cluster data
c) Classify categories
d) Build rules
Answer: a
155. Clustering is a type of:
a) Supervised learning
b) Unsupervised learning
c) Reinforcement learning
d) Rule-based AI
Answer: b
156. K-Means is an example of:
a) Clustering
b) Regression
c) Classification
d) Reinforcement learning
Answer: a
157. Decision trees are used for:
a) Classification and regression
b) Clustering only
c) Rule-based AI
d) Reinforcement learning only
Answer: a
158. Random Forest is:
a) Ensemble of decision trees
b) Single decision tree
c) Neural network
d) Clustering algorithm
Answer: a
159. SVM stands for:
a) Support Vector Machine
b) Supervised Vector Model
c) Simple Variable Model
d) Sequential Vector Machine
Answer: a
160. Naive Bayes is based on:
a) Probability theory
b) Neural networks
c) Rules only
d) Clustering
Answer: a
161. Linear regression predicts:
a) Continuous values
b) Categories
c) Clusters
d) Rules
Answer: a
162. Logistic regression predicts:
a) Probabilities for categorical outcomes
b) Continuous values
c) Clusters
d) Neural outputs
Answer: a
163. Feature scaling is important because:
a) Algorithms are sensitive to magnitude
b) Improves data visualization
c) Reduces storage
d) None
Answer: a
164. Standardization means:
a) Scaling data to zero mean and unit variance
 b) Normalizing data between 0 and 1
c) Reducing features
d) Selecting features
Answer: a
165. Normalization means:
a) Scaling data between 0 and 1
b) Zero mean and unit variance
c) Feature selection
d) Clustering data
Answer: a
166. Overfitting occurs when:
a) Model performs well on training but poorly on test data
b) Model underfits
c) Model has fewer features
d) Model generalizes well
Answer: a
167. Underfitting occurs when:
a) Model cannot capture underlying patterns
b) Model memorizes data
c) Model generalizes well
d) Model performs perfectly
Answer: a
168. Cross-validation is used to:
a) Evaluate model performance
b) Scale features
c) Reduce dimensionality
d) Preprocess data
Answer: a
169. K-Fold cross-validation splits data into:
a) K subsets
b) One subset
c) Two subsets
d) Ten only
Answer: a
170. Bias refers to:
a) Error due to assumptions in the model
b) Random noise
c) Overfitting
d) Variance
Answer: a
171. Variance refers to:
a) Error due to sensitivity to data fluctuations
b) Error due to assumptions
c) Training accuracy
d) Test accuracy
Answer: a
172. Bias-variance tradeoff aims to:
a) Balance underfitting and overfitting
b) Increase bias
c) Increase variance
 d) Reduce accuracy
Answer: a
173. Dimensionality reduction helps in:
a) Reducing number of features
b) Increasing noise
c) Making data complex
d) Overfitting
Answer: a
174. PCA stands for:
a) Principal Component Analysis
b) Predictive Clustering Algorithm
c) Probabilistic Classification Analysis
d) Pattern Component Analysis
Answer: a
175. KNN stands for:
a) K-Nearest Neighbors
b) Kernel Neural Network
c) Knowledge Neural Net
d) K-Node Network
Answer: a
176. KNN is used for:
a) Classification and regression
b) Clustering
c) Dimensionality reduction
d) Feature extraction
Answer: a
177. In KNN, distance metrics include:
a) Euclidean and Manhattan
b) Minkowski only
c) Cosine similarity only
d) None
Answer: a
178. Hierarchical clustering forms:
a) Tree-like structure
b) Flat clusters only
c) Rules
d) Regression models
Answer: a
179. DBSCAN is:
a) Density-based clustering algorithm
b) Regression algorithm
c) Classification algorithm
d) Decision tree variant
Answer: a
180. Silhouette score evaluates:
a) Quality of clustering
b) Classification accuracy
c) Regression error
d) Feature importance
Answer: a
181. Confusion matrix is used in:
a) Classification evaluation
b) Regression evaluation
c) Clustering evaluation
d) Feature selection
Answer: a
182. Precision measures:
a) Correct positive predictions / total predicted positives
b) Correct negatives / total negatives
c) Overall accuracy
d) Training error
Answer: a
183. Recall measures:
a) Correct positive predictions / total actual positives
b) Correct negatives / total negatives
c) Overall accuracy
d) Feature importance
Answer: a
184. F1-score is:
a) Harmonic mean of precision and recall
b) Average of precision and recall
c) Difference of precision and recall
d) Ratio of precision to recall
Answer: a
185. ROC curve plots:
a) True positive rate vs False positive rate
b) Precision vs Recall
c) Accuracy vs Error
d) Variance vs Bias
Answer: a
186. AUC stands for:
a) Area Under Curve
b) Accuracy Under Classification
c) Algorithmic Utility Coefficient
d) None
Answer: a
187. Feature importance can be computed using:
a) Random Forest
b) Decision Tree
c) Gradient Boosting
d) All of the above
Answer: d
188. Gradient Descent is used for:
a) Optimizing model parameters
b) Evaluating clustering
c) Feature scaling
d) Data normalization
Answer: a
189. Learning rate affects:
a) Step size in gradient descent
 b) Accuracy only
c) Feature importance
d) Clustering quality
Answer: a
190. Epochs refer to:
a) Number of times entire dataset is used to train the model
b) Features selected
c) Number of clusters
d) Decision rules applied
Answer: a
191. Batch size refers to:
a) Number of samples processed at a time during training
b) Total dataset size
c) Number of features
d) Number of clusters
Answer: a
192. Early stopping helps to:
a) Prevent overfitting
b) Prevent underfitting
c) Increase bias
d) Reduce variance
Answer: a
193. Regularization helps to:
a) Reduce overfitting
b) Increase model complexity
c) Remove features
d) Normalize data
Answer: a
194. L1 regularization is also called:
a) Lasso
b) Ridge
c) Elastic Net
d) Dropout
Answer: a
195. L2 regularization is also called:
a) Ridge
b) Lasso
c) Elastic Net
d) Batch normalization
Answer: a
196. Elastic Net combines:
a) L1 and L2 regularization
b) L1 only
c) L2 only
d) None
Answer: a
197. Outlier detection helps in:
a) Removing anomalies that affect model
b) Clustering only
c) Classification only
 d) Regression only
Answer: a
198. Z-score is used for:
a) Standardizing data
b) Normalizing data
c) Clustering
d) Decision trees
Answer: a
199. PCA reduces:
a) Dimensionality of data
b) Sample size
c) Clusters
d) Features only
Answer: a
200. Scree plot is used in:
a) PCA to determine number of components
b) Clustering evaluation
c) Regression analysis
d) Neural network optimization
Answer: a

Section 5: Applications of AI (201–250 MCQs)

201. AI is widely used in healthcare for:
a) Disease prediction
b) Medical image analysis
c) Personalized treatment
d) All of the above
Answer: d
202. AI-powered chatbots are used for:
a) Customer support
b) Automated responses
c) Virtual assistants
d) All of the above
Answer: d
203. Autonomous vehicles use AI for:
a) Navigation
b) Obstacle detection
c) Traffic prediction
d) All of the above
Answer: d
204. AI in finance can be used for:
a) Fraud detection
b) Algorithmic trading
c) Risk assessment
d) All of the above
Answer: d
205. Recommendation systems use AI for:
a) Suggesting products or services
b) Clustering users
 c) Predicting preferences
d) All of the above
Answer: d
206. AI in agriculture helps with:
a) Crop disease detection
b) Yield prediction
c) Soil analysis
d) All of the above
Answer: d
207. AI in manufacturing is used for:
a) Predictive maintenance
b) Quality control
c) Process optimization
d) All of the above
Answer: d
208. NLP is used in AI applications for:
a) Text analysis
b) Sentiment analysis
c) Language translation
d) All of the above
Answer: d
209. Image recognition in AI is used for:
a) Object detection
b) Facial recognition
c) Medical imaging
d) All of the above
Answer: d
210. Speech recognition systems use AI to:
a) Convert speech to text
b) Analyze tone and emotion
c) Control devices via voice
d) All of the above
Answer: d
211. Virtual assistants like Alexa and Siri are examples of:
a) NLP applications
b) ML applications
c) AI-powered systems
d) All of the above
Answer: d
212. Fraud detection systems in banking use AI to:
a) Identify suspicious transactions
b) Predict risky customers
c) Improve security
d) All of the above
Answer: d
213. AI in healthcare imaging includes:
a) MRI analysis
b) CT scan interpretation
c) X-ray diagnosis
 d) All of the above
Answer: d
214. AI in retail helps with:
a) Inventory management
b) Customer recommendations
c) Price optimization
d) All of the above
Answer: d
215. Predictive maintenance in manufacturing uses AI to:
a) Detect machine failures before they happen
b) Schedule repairs efficiently
c) Optimize operations
d) All of the above
Answer: d
216. AI in transportation can optimize:
a) Traffic flow
b) Route planning
c) Fuel consumption
d) All of the above
Answer: d
217. AI-powered drones are used for:
a) Surveillance
b) Delivery
c) Agricultural monitoring
d) All of the above
Answer: d
218. AI in cybersecurity helps with:
a) Threat detection
b) Malware analysis
c) Network monitoring
d) All of the above
Answer: d
219. AI in education can be used for:
a) Personalized learning
b) Automated grading
c) Student performance prediction
d) All of the above
Answer: d
220. AI-powered translation systems like Google Translate use:
a) NLP
b) Machine Learning
c) Deep Learning
d) All of the above
Answer: d
221. Autonomous cars use AI for:
a) Sensor fusion
b) Decision-making
c) Path planning
d) All of the above
Answer: d
222. AI in robotics enables:
a) Object manipulation
b) Path planning
c) Human-robot interaction
d) All of the above
Answer: d
223. AI in smart homes controls:
a) Lighting
b) Temperature
c) Security
d) All of the above
Answer: d
224. AI in social media is used for:
a) Content recommendation
b) Moderation of harmful content
c) Targeted advertising
d) All of the above
Answer: d
225. AI in energy management helps:
a) Optimize consumption
b) Predict demand
c) Integrate renewable sources
d) All of the above
Answer: d
226. AI in marketing helps with:
a) Customer segmentation
b) Campaign optimization
c) Personalized promotions
d) All of the above
Answer: d
227. AI-powered search engines improve:
a) Query understanding
b) Ranking results
c) User experience
d) All of the above
Answer: d
228. AI in gaming is used for:
a) NPC behavior
b) Procedural content generation
c) Game testing
d) All of the above
Answer: d
229. AI in law enforcement is applied for:
a) Predictive policing
b) Crime pattern analysis
c) Surveillance
d) All of the above
Answer: d
230. AI in weather forecasting uses:
a) Historical data
 b) Sensor data
c) Machine learning models
d) All of the above
Answer: d
231. AI in supply chain management helps with:
a) Demand forecasting
b) Inventory optimization
c) Logistics planning
d) All of the above
Answer: d
232. AI in autonomous drones helps with:
a) Object detection
b) Path optimization
c) Real-time decision-making
d) All of the above
Answer: d
233. AI in healthcare chatbots can:
a) Provide symptom checks
b) Answer patient queries
c) Schedule appointments
d) All of the above
Answer: d
234. AI in fraud prevention can:
a) Detect anomalies
b) Block suspicious transactions
c) Predict fraudulent patterns
d) All of the above
Answer: d
235. AI-powered financial advisors are called:
a) Robo-advisors
b) Expert systems
c) Decision trees
d) Neural networks
Answer: a
236. AI in smart cities helps with:
a) Traffic management
b) Energy distribution
c) Public safety
d) All of the above
Answer: d
237. AI in healthcare can improve:
a) Diagnosis accuracy
b) Treatment planning
c) Patient outcomes
d) All of the above
Answer: d
238. AI in e-commerce helps with:
a) Product recommendations
b) Inventory management
c) Dynamic pricing
 d) All of the above
Answer: d
239. AI in autonomous navigation uses:
a) Sensor data
b) ML models
c) Path planning algorithms
d) All of the above
Answer: d
240. AI in fraud detection uses:
a) Anomaly detection
b) Pattern recognition
c) Risk scoring
d) All of the above
Answer: d
241. AI in marketing analytics helps:
a) Identify customer trends
b) Predict sales
c) Optimize campaigns
d) All of the above
Answer: d
242. AI in natural disaster management helps with:
a) Early warning
b) Damage prediction
c) Resource allocation
d) All of the above
Answer: d
243. AI in predictive maintenance reduces:
a) Downtime
b) Operational costs
c) Unplanned failures
d) All of the above
Answer: d
244. AI in sports analytics can:
a) Track player performance
b) Predict match outcomes
c) Optimize training
d) All of the above
Answer: d
245. AI in HR management helps with:
a) Resume screening
b) Candidate matching
c) Employee performance tracking
d) All of the above
Answer: d
246. AI in personalized medicine uses:
a) Patient data
b) Genomic data
c) Machine learning models
d) All of the above
Answer: d
247. AI in autonomous drones for agriculture:
a) Monitors crop health
b) Sprays fertilizers
c) Predicts yield
d) All of the above
Answer: d
248. AI in traffic management can:
a) Reduce congestion
b) Optimize signal timing
c) Predict accidents
d) All of the above
Answer: d
249. AI in education helps:
a) Adaptive learning
b) Grading automation
c) Performance prediction
d) All of the above
Answer: d
250. AI in environmental monitoring can:
a) Track pollution levels
b) Monitor wildlife
c) Predict natural hazards
d) All of the above
Answer: d
251. • Reinforcement learning is a type of:
a) Supervised learning
b) Unsupervised learning
c) Trial-and-error learning
d) Rule-based AI
Answer: c
252. • In reinforcement learning, the agent learns by:
a) Receiving rewards or penalties
b) Following fixed rules
c) Using labeled data
d) Clustering inputs
Answer: a
253. • Q-Learning is used in:
a) Reinforcement learning
b) Supervised learning
c) Unsupervised learning
d) Rule-based systems
Answer: a
254. • Deep reinforcement learning combines:
a) Neural networks and reinforcement learning
b) Decision trees and supervised learning
c) Clustering and regression
d) Rule-based AI and ML
Answer: a
255. • Overfitting occurs when a model:
a) Memorizes training data and performs poorly on new data
 b) Generalizes well
c) Has low variance
d) None of the above
Answer: a
256. • Underfitting occurs when a model:
a) Cannot capture underlying patterns
b) Memorizes training data
c) Generalizes perfectly
d) Is overtrained
Answer: a
257. • Bagging improves model performance by:
a) Combining predictions of multiple models
b) Reducing features
c) Scaling data
d) Using a single model only
Answer: a
258. • Boosting improves model performance by:
a) Sequentially correcting errors of previous models
b) Reducing features
c) Using a single model
d) Clustering data
Answer: a
259. • Random Forest is an example of:
a) Bagging ensemble
b) Boosting ensemble
c) Single decision tree
d) KNN algorithm
Answer: a
260. • Gradient Boosting is an example of:
a) Boosting ensemble
b) Bagging ensemble
c) Single tree
d) Clustering algorithm
Answer: a
261. • Hyperparameters are:
a) Settings of the model defined before training
b) Model outputs
c) Feature values
d) Data samples
Answer: a
262. • Model parameters are:
a) Learned during training
b) Predefined settings
c) Features
d) Hyperparameters
Answer: a
263. • Cross-entropy loss is used for:
a) Classification tasks
b) Regression tasks
c) Clustering tasks
 d) Feature scaling
Answer: a
264. • Mean squared error (MSE) is used for:
a) Regression tasks
b) Classification tasks
c) Clustering tasks
d) Dimensionality reduction
Answer: a
265. • Softmax function is used to:
a) Convert logits into probabilities
b) Calculate distances
c) Scale features
d) Reduce variance
Answer: a
266. • ReLU activation stands for:
a) Rectified Linear Unit
b) Regularized Linear Unit
c) Random Linear Unit
d) Reinforced Learning Unit
Answer: a
267. • Sigmoid activation is used for:
a) Probabilities in binary classification
b) Regression outputs
c) Clustering
d) Decision trees
Answer: a
268. • Tanh activation outputs range:
a) -1 to 1
b) 0 to 1
c) -∞ to ∞
d) 0 to ∞
Answer: a
269. • Dropout in neural networks helps to:
a) Prevent overfitting
b) Increase training speed
c) Reduce features
d) Normalize data
Answer: a
270. • Batch normalization helps to:
a) Stabilize and accelerate training
b) Reduce features
c) Increase learning rate
d) Reduce dataset size
Answer: a
271. • Convolutional Neural Networks (CNNs) are used for:
a) Image recognition
b) Text classification
c) Regression
d) Clustering
Answer: a
272. • Recurrent Neural Networks (RNNs) are used for:
a) Sequential data processing
b) Image classification only
c) Clustering only
d) Decision trees
Answer: a
273. • LSTM networks improve RNNs by:
a) Handling long-term dependencies
b) Reducing number of layers
c) Scaling features
d) Reducing variance
Answer: a
274. • Autoencoders are used for:
a) Dimensionality reduction and feature learning
b) Clustering only
c) Regression only
d) Classification only
Answer: a
275. • Generative Adversarial Networks (GANs) consist of:
a) Generator and discriminator networks
b) Input and output layers only
c) Decision tree ensembles
d) Feature selection models
Answer: a
276. • K-Means clustering requires:
a) Number of clusters (k)
b) Labels
c) Regression targets
d) Decision rules
Answer: a
277. • Silhouette score ranges from:
a) -1 to 1
b) 0 to 100
c) -∞ to ∞
d) None
Answer: a
278. • Dimensionality reduction helps to:
a) Reduce computation cost
b) Remove noise
c) Visualize high-dimensional data
d) All of the above
Answer: d
279. • PCA components are:
a) Orthogonal and capture maximum variance
b) Random projections
c) Clusters
d) Regression targets
Answer: a
280. • t-SNE is used for:
a) Non-linear dimensionality reduction
b) Clustering only
c) Regression only
d) Classification only
Answer: a
281. • Feature engineering improves model performance by:
a) Creating meaningful input features
b) Removing outliers only
c) Scaling features only
d) Selecting hyperparameters
Answer: a
282. • One-hot encoding is used for:
a) Categorical variables
b) Continuous variables
c) Labels only
d) Features only
Answer: a
283. • Label encoding is used for:
a) Converting categories into numeric labels
b) Continuous variables
c) Regression targets only
d) Clustering features
Answer: a
284. • SMOTE is used for:
a) Balancing imbalanced datasets
b) Dimensionality reduction
c) Scaling features
d) Clustering
Answer: a
285. • ROC-AUC evaluates:
a) Classification performance
b) Regression accuracy
c) Clustering quality
d) Dimensionality reduction
Answer: a
286. • Hyperparameter tuning can be done via:
a) Grid search or random search
b) Cross-validation only
c) Feature scaling
d) Clustering
Answer: a
287. • Early stopping prevents:
a) Overfitting
b) Underfitting
c) Bias
d) Variance
Answer: a
288. • Ensemble learning improves:
a) Accuracy and robustness
b) Dataset size
c) Feature scaling
d) Model simplicity
Answer: a
289. • Bagging reduces:
a) Variance
b) Bias
c) Noise
d) Features
Answer: a
290. • Boosting reduces:
a) Bias
b) Variance
c) Noise
d) Features
Answer: a
291. • Stacking combines:
a) Multiple base learners using a meta-learner
b) Features only
c) Data preprocessing
d) Scaling methods
Answer: a
292. • AI explainability helps to:
a) Understand model decisions
b) Scale features
c) Optimize hyperparameters
d) Reduce noise
Answer: a
293. • Shapley values are used for:
a) Feature contribution analysis
b) Outlier detection
c) Clustering evaluation
d) Regression
Answer: a
294. • LIME helps to:
a) Explain predictions of complex models
b) Reduce features
c) Optimize parameters
d) Normalize data
Answer: a
295. • AI ethics considers:
a) Bias and fairness
b) Transparency
c) Accountability
d) All of the above
Answer: d
296. • Transfer learning allows:
a) Using pre-trained models for new tasks
b) Training from scratch only
c) Clustering only
d) Regression only
Answer: a
297. • Fine-tuning in transfer learning:
a) Adjusts pre-trained models on new data
b) Trains from scratch
c) Reduces features
d) Normalizes data
Answer: a
298. • Reinforcement learning applications include:
a) Robotics
b) Game AI
c) Autonomous vehicles
d) All of the above
Answer: d
299. • AI in healthcare prediction uses:
a) ML models for disease outcomes
b) Data-driven insights
c) Personalized treatment
d) All of the above
Answer: d
300. • The main goal of AI/ML in real-world applications is to:
a) Automate tasks and improve decision-making
b) Replace humans entirely
c) Only predict numbers
d) Only classify images
Answer: a