Analysis of AI Exam Question Patterns:
Predictions for Upcoming Exams
This analysis examines previous year question papers from 2023-2024 in Artificial Intelligence
courses to identify patterns, topic emphasis, and predict likely questions for future exams.
Key Pattern Findings from PYQ Analysis
The analysis of four examination papers from 2023-2024 reveals clear patterns in question
distribution, repetition, and topic emphasis. While no exact duplicate questions were found,
there were several high-similarity question pairs with considerable conceptual overlap.
Topic Distribution Analysis
The frequency analysis shows consistent emphasis on core AI concepts across both years, with
notable shifts in focus:
2023 Topic Distribution:
Heuristic approaches (3 questions)
Search algorithms including water jug problems (2)
A* search implementation (2)
Minimax and alpha-beta pruning (2 each)
Constraint satisfaction problems (2)
Various classical AI problems (missionaries/cannibals, 8-puzzle)
2024 Topic Distribution:
Constraint satisfaction techniques (2 questions)
8-puzzle implementations (2)
Heuristic functions (2)
Game theory approaches (minimax)
Classical AI problems with less emphasis on water jug problems
This distribution highlights a consistent core curriculum while showing subtle shifts toward more
application-based questions in recent papers.
Exact Repeated Questions from Past Years
The following questions appeared with minimal variation across multiple papers:
1. Water Jug Problem Implementation
2023: "Apply Graph version of Depth first search on Three water jug problem [capacity
(8, 5, 3), start state (8, 0, 0), goal state (4, 4, 0)]" [1]
2023 (Ms): "Apply Graph version of Breadth first search on Three water jug problem
[capacity (8, 5, 3), start state (8, 0, 0), goal state (4, 4, 0)]" [2]
2. Alpha-Beta Pruning Application
2023: "Apply minimax algorithm and Alpha Beta pruning on the following Figure to find
out value of root node. How may evaluations are saved while applying Alpha Beta
pruning?" [1]
2023 (Ms): Nearly identical question with the same diagram and wording [2]
2024: Similar question focusing on Alpha-Beta pruning application [3]
3. 8-Puzzle Problem Analysis
2023: Properties of 8-puzzle state spaces [1]
2024: "Apply Heuristic function to solve the above 8 puzzle problem" [3]
2024 (MS): "Apply A* Algorithm with Manhattan Heuristic function" for 8-puzzle [4]
Conceptually Similar Questions with Variations
Several question types appear consistently but with modified contexts or parameters:
1. State Space Problem Representations
The exams consistently feature questions on state space representation, but with different
problem domains:
Missionaries and cannibals problem (2023) [1]
Three jealous couples problem (2024) [3]
Tower of Hanoi problem (2023 Ms) [2]
These questions typically ask for:
State representation formulation
Search tree visualization
Application of specific search algorithms (BFS/DFS)
2. Cryptarithmetic Problems as CSPs
Consistent appearance with different letter combinations:
2024: "CROSS + ROADS = DANGER" [3]
2024 (MS): "LOGIC + LOGIC = PROLOG" [4]
3. First-Order Logic Translations
Regular appearance of English-to-FOL translation questions with different sentences:
2023: Translation tasks with quantified statements [1]
2024: Multiple statements about surgeons, lawyers, etc. [4]
2023 (Ms): Sentences about John, food, and eating habits [2]
4. Bayesian Probability Applications
2023: Naïve Bayes application on general dataset [1]
2024: Car theft prediction using Naïve Bayes [3]
Emerging Trends in Question Patterns
Based on the year-over-year analysis, several trends in question formulation are becoming
apparent:
1. Increased Emphasis on Applied Heuristics
More questions are focusing on the practical application of heuristic functions rather than
theoretical properties:
A* with Manhattan distance for pathfinding [4]
Heuristic evaluation in game trees
2. Shift Toward Real-world Applications
Recent papers show more questions framed around practical scenarios:
Car theft prediction using Naïve Bayes (2024) [3]
Class scheduling as a constraint satisfaction problem [1]
3. Integration of Multiple Concepts in Single Questions
Questions increasingly require applying multiple AI techniques together:
Combining search algorithms with specific heuristics
Applying pruning techniques to game trees with evaluation functions
Highest Probability Questions for Upcoming Exams
Based on statistical analysis and pattern recognition, these questions have the highest
probability of appearing (or appearing with slight modifications) in upcoming exams:
1. Water Jug Problem (Very High Probability)
Likely to appear with either BFS or DFS implementation requirement
Specific capacities may change, but structure will remain
2. 8-Puzzle with A and Manhattan Distance* (Very High Probability)
Consistently appears across years with high weighting
Initial/goal states will change but concept remains constant
3. Minimax with Alpha-Beta Pruning (High Probability)
Tree structure may change, but question format is consistent across years
Often asks for both root node value and pruning efficiency
4. Constraint Satisfaction for Cryptarithmetic (High Probability)
Different letter combinations, same fundamental question structure
May include additional constraints or variations
5. First-Order Logic Translation and Resolution (Medium-High Probability)
Translation tasks with different sentences but similar complexity
Resolution proof requirements based on given axioms
6. Naïve Bayes Application to Classification Problems (Medium Probability)
Likely to feature a new dataset with similar structure
May incorporate additional probability calculation requirements
Conclusion
The analysis reveals strong consistency in core topics across examination papers while showing
evolution in application contexts and complexity. Students preparing for upcoming exams
should focus on mastering the high-probability topics identified above, particularly classical
search problems, constraint satisfaction techniques, and game theory applications, as these
form the foundation of the assessment pattern.
The shift toward more application-oriented questions suggests that understanding the practical
implementation of algorithms, rather than just theoretical knowledge, will be increasingly
important in future examinations.
⁂
1. https://ppl-ai-file-upload.s3.amazonaws.com/web/direct-files/attachments/62225915/c47457d0-3dab-
4a4f-8834-92aabf169bf9/ARTIFICIAL-INTELLIGENCE-PRINCIPLES-s23.pdf
2. https://ppl-ai-file-upload.s3.amazonaws.com/web/direct-files/attachments/62225915/d4dd426c-d44f-
49de-95c3-72f89b3083c3/ARTIFICIAL-INTELLIGENCE-PRINCIPLES-AND-TECHNIQUES-Ms23.pdf
3. https://ppl-ai-file-upload.s3.amazonaws.com/web/direct-files/attachments/62225915/007cc328-b8f5-
499d-978d-036a90babb26/ARTIFICIAL-INTELLIGENCE-PRINCIPLES-AND-TECHNIQUES-s24.pdf
4. https://ppl-ai-file-upload.s3.amazonaws.com/web/direct-files/attachments/62225915/fc775589-6f80-
4a0e-8e74-f45694a6f37c/ARTIFICIAL-INTELLIGENCE-MS24.pdf
