AL-304 Artificial Intelligence / Introduction to Artificial Intelligence - Jun 2024 Question Paper

Under each question, How to answer lists the points an answer should cover, outlined by CampusPrep from the syllabus. Check the details against your textbook.

1. 7 Marks
   aWhat are the essential characteristics of production systems and how do they differ from other AI problem-solving approaches?
   How to answer

   • Define a production system: set of rules (condition-action), working memory, and control strategy.
   • List key characteristics: modularity, uniformity, naturalness, separation of knowledge and control.
   • Contrast with other paradigms (e.g., state-space search, imperative algorithms, connectionist models).
   • Discuss advantages and typical use cases in AI.
2. 7 Marks
   bWhat is the $A^*$ search algorithm, and how does it combine elements of other search algorithms? Why is it considered a popular choice for many AI applications?
   How to answer

   • Define $A^*$ algorithm and its evaluation function: $f(n) = g(n) + h(n)$.
   • Explain combination of Uniform Cost Search / Dijkstra ($g(n)$) and Greedy Best-First ($h(n)$).
   • State properties: completeness and optimality (under admissible and consistent heuristics).
   • Explain popularity: efficiency, guarantees shortest path, widely applicable in pathfinding and planning.
3. 7 Marks
   aHow can AI contribute to solving global challenges, such as climate change, healthcare or education? Provide specific examples.
   How to answer

   • Introduction to AI's impact on global socio-economic and environmental challenges.
   • Climate change: renewable energy grid optimization, carbon tracking, climate prediction models.
   • Healthcare: medical imaging diagnosis, drug discovery, personalized medicine.
   • Education: intelligent tutoring systems, automated grading, personalized learning pathways.
4. 7 Marks
   bHow does knowledge representation facilitate the process of deductive reasoning in AI?
   How to answer

   • Define knowledge representation (KR) and deductive reasoning.
   • Explain how formal KR schemes (like propositional and first-order logic) provide syntax and semantics.
   • Describe inference engines applying deductive rules (e.g., Modus Ponens, resolution) over KR structures.
   • Highlight how structured representation ensures sound and complete derivation of new truths.
5. 7 Marks
   aCompare and contrast propositional logic and predicate logic in terms of their expressiveness and applicability in different problem domains.
   How to answer

   • Define Propositional Logic (facts as atomic propositions) vs Predicate Logic (objects, predicates, functions, quantifiers $\forall, \exists$).
   • Compare expressiveness: Predicate logic can represent generalized statements and relationships; propositional logic cannot.
   • Compare computational complexity and decidability.
   • Provide domain examples illustrating the need for predicate logic over propositional logic.
6. 7 Marks
   bAnalyze a complex AI problem, such as natural language understanding and propose a knowledge representation scheme that would effectively address its challenges.
   How to answer

   • Analyze core challenges of NLU: ambiguity (lexical, syntactic, semantic), context-dependency, pragmatics.
   • Propose an appropriate KR scheme (e.g., Conceptual Dependency, Ontologies, or Knowledge Graphs).
   • Explain how the chosen scheme models entities, relationships, context, and inference.
   • Discuss advantages and limitations of the proposed approach.
7. 7 Marks
   aDifferentiate between scripts, schemas and frames. How are they applied in knowledge representation?
   How to answer

   • Define frames: slot-and-filler structures representing objects, concepts, and default values.
   • Define scripts: specialized frame-like structures describing stereotypical sequences of events (e.g., restaurant script).
   • Define schemas: generalized mental or computational frameworks for organizing concepts.
   • Tabulate differences (focus, structure, temporal ordering).
   • Explain practical applications of each in KR and reasoning.
8. 7 Marks
   bDefine Bayes' Theorem and explain its significance in probabilistic reasoning.
   How to answer

   • State Bayes' Theorem mathematically: $P(A|B) = \frac{P(B|A)P(A)}{P(B)}$.
   • Define terms: prior probability $P(A)$, posterior $P(A|B)$, likelihood $P(B|A)$, and marginal evidence $P(B)$.
   • Explain handling of uncertainty and belief updating in AI.
   • Discuss significance in Bayesian networks, medical diagnostics, and spam filtering.
9. 7 Marks
   aEvaluate the limitations of frames in knowledge representation and propose enhancements or alternatives to address these limitations.
   How to answer

   • Identify limitations of frame systems: exception handling, multiple inheritance anomalies, lack of formal semantics, rigidity.
   • Propose enhancements: default overriding rules, non-monotonic reasoning mechanisms, constraints on slots.
   • Discuss alternatives: Description Logics, OWL, semantic web ontologies, object-oriented KR models.
10. 7 Marks
    bWhat is the "block world" problem in robotics and why is it a significant challenge in the field of artificial intelligence?
    How to answer

    • Define the blocks world domain: table, set of blocks, robotic arm with operations (stack, unstack, pickup, putdown).
    • Describe predicates used (e.g., ON, ONTABLE, CLEAR, HOLDING, ARMEMPTY).
    • Explain why it is significant: classic benchmark for automated planning, reasoning about actions, and goal interactions (e.g., Sussman anomaly).
    • Highlight its foundational role in STRIPS and modern planning algorithms.
11. 7 Marks
    aGiven a simplified game tree for tic-tac-toe, apply the minimax procedure to determine the optimal move for a player at a particular state of the game.
    How to answer

    • Draw or specify a simplified game tree for a tic-tac-toe partial board state.
    • Define MAX and MIN player roles and utility values (+1, 0, -1).
    • Demonstrate bottom-up propagation of values using Minimax rules.
    • Identify and justify the optimal choice for the current player at the root node.
12. 7 Marks
    bEvaluate the ethical implications of using NLP in surveillance and monitoring systems. How can these implications be addressed responsibly?
    How to answer

    • Discuss ethical issues: invasion of privacy, mass surveillance, speech censorship, false positives/negatives.
    • Analyze bias in NLP models leading to discrimination against linguistic/demographic groups.
    • Propose technical and policy solutions: differential privacy, anonymization, explainability.
    • Discuss regulatory frameworks, ethical guidelines, and legal oversight.
13. 7 Marks
    aList and describe the key characteristics that distinguish expert systems from other computer programs.
    How to answer

    • Define an Expert System.
    • List key differentiating features: explicit separation of knowledge base and inference engine, explanation facility (answering 'why' and 'how'), symbolic reasoning, tolerance of error/uncertainty.
    • Compare with conventional programs (algorithmic, hard-coded data and logic, deterministic output).
    • Conclude with real-world examples (e.g., MYCIN, DENDRAL).
14. 7 Marks
    bDevelop a comprehensive expert system for a complex domain, such as legal advice or financial planning. Explain how you would ensure the system's knowledge is up to date.
    How to answer

    • Select a domain (e.g., Legal Advice or Financial Planning).
    • Describe system architecture: Knowledge Base, Inference Engine, Working Memory, User Interface, Explanation Module.
    • Detail the knowledge acquisition process from human experts and regulations.
    • Present mechanisms to ensure up-to-date knowledge: dynamic rule updating, integration with live feeds, automated knowledge refinement, continuous validation by domain experts.
15. 14 Marks
    Attempt any two: a) Analyze the limitations of expert systems, including issues related to knowledge acquisition, knowledge representation. b) Differentiate between classical search algorithms and heuristic search algorithms c) Compare and contrast classical AI systems and modern AI systems in terms of their performance and adaptability. d) Define natural language processing (NLP) and its importance in AI.
    How to answer

    • For (a): Explain knowledge acquisition bottleneck, brittleness, lack of common sense, maintenance difficulties.
    • For (b): Compare uninformed/blind search (BFS, DFS) vs informed/heuristic search (A*, Greedy Best-First) based on domain knowledge and time/space complexity.
    • For (c): Contrast rule-based/symbolic AI (deterministic, rigid, non-adaptive) with data-driven/modern AI (probabilistic, deep learning, highly adaptable).
    • For (d): Define NLP, outline core components (NLU, NLG), and explain importance in bridging human-machine communication.