(SEM VII) THEORY EXAMINATION 2023-24 ARTIFICIAL INTELLIGENCE

SECTION A (2 Marks Each – Short Answers)

a) Historical background and evolution of AI

Artificial Intelligence originated in the 1950s with the goal of making machines think like humans. The term “Artificial Intelligence” was coined by John McCarthy in 1956. Early AI focused on problem-solving and symbolic reasoning. Over time, AI evolved through expert systems, machine learning, and today includes deep learning, NLP, and autonomous systems.

b) Definition of AI and its main objectives

Artificial Intelligence is a branch of computer science that focuses on creating systems capable of performing tasks that require human intelligence.
 

Objectives: learning, reasoning, problem-solving, perception, and decision-making.

c) Challenges with partial observations in search problems

In partial observation environments, the agent does not have complete information about the current state. This leads to uncertainty, incorrect decisions, increased search complexity, and difficulty in selecting optimal actions.

d) Constraint Satisfaction Problems (CSP)

A Constraint Satisfaction Problem is defined by a set of variables, their domains, and constraints that restrict the values variables can take. The goal is to find values for all variables that satisfy all constraints.

e) Unification in logic programming

Unification is the process of making two logical expressions identical by finding suitable substitutions for variables. It is a key operation in Prolog for pattern matching.

f) Resolution in logic programming

Resolution is an inference technique used in logic programming and automated theorem proving. It derives new clauses by eliminating complementary literals from two clauses.

g) Characteristics of an intelligent agent in a multi-agent system

Key characteristics include autonomy, social ability, reactivity, proactiveness, and the ability to cooperate or compete with other agents.

h) Importance of communication among intelligent agents

Communication allows agents to share knowledge, coordinate actions, negotiate, and resolve conflicts, improving overall system efficiency.

i) Real-world applications of information extraction

Examples include extracting customer details from forms, medical data from reports, financial data from documents, and news analysis.

j) Challenges in information retrieval from large datasets

Challenges include data volume, unstructured formats, ambiguity in queries, relevance ranking, and maintaining retrieval accuracy.

SECTION B (10 Marks – Long Answers)

a) Role of sensors and effectors in intelligent agents

Sensors allow agents to perceive the environment by collecting data such as images, sound, or signals. Effectors enable agents to act upon the environment, such as motors or displays. Together, they form the perception–action cycle essential for intelligent behavior.

b) Uninformed search strategies

Uninformed search strategies do not use domain knowledge. They explore the search space systematically.
 

Examples:

Breadth First Search (BFS)

Depth First Search (DFS)

Uniform Cost Search

These methods are simple but may be inefficient for large problems.

c) First Order Predicate Logic and Prolog

First Order Predicate Logic (FOPL) represents facts and relationships using predicates, variables, and quantifiers. Prolog uses FOPL to perform logical inference through facts, rules, and queries using unification and backtracking.

d) Perception and action in multi-agent systems

Agents perceive the environment using sensors, process information individually or collectively, and act through effectors. Interaction among agents enables cooperation, coordination, and competition within shared environments.

e) Importance of pre-trained language models

Pre-trained language models learn linguistic patterns from large datasets. They improve tasks like translation, sentiment analysis, chatbots, and question answering while reducing training time and resource requirements.

SECTION C (10 Marks – Any One)

3(a) AI problem-solving using search and heuristics

AI systems solve problems by exploring possible states using search algorithms. Heuristics guide the search toward promising paths, reducing time and computational cost. Algorithms like A* combine search with heuristics for efficiency.

3(b) Ethical considerations in AI

Ethical issues include data privacy, bias, transparency, accountability, job displacement, and misuse of AI. Responsible AI development ensures fairness, safety, and societal benefit.

SECTION D / E / F / G (10 Marks – Key Answers)

Local search algorithms

Local search algorithms work by iteratively improving a solution based on neighboring states.
 

Example: Hill Climbing used in optimization problems like scheduling.

Informed search and heuristics

Informed search uses heuristics to estimate the cost to reach the goal. Heuristics improve efficiency by reducing unnecessary exploration.

Forward vs Backward chaining

Forward chaining is data-driven, starting from facts to conclusions.
Backward chaining is goal-driven, starting from goals to known facts.

Ontological engineering

It represents knowledge using concepts, relationships, and rules. It helps build intelligent systems by enabling shared understanding and reasoning.

Communication paradigms in intelligent agents

Agents communicate using direct messaging, blackboard systems, and negotiation protocols to collaborate effectively.

Bargaining among intelligent agents

Bargaining helps agents resolve conflicts by negotiating terms to reach mutually beneficial agreements.

Language models in NLP

Language models predict word sequences and enable tasks like text generation, summarization, translation, and sentiment analysis.

Role of information retrieval

Information retrieval enhances search engines and recommendation systems by efficiently finding relevant content from large datasets.