←Back to Notes

AI • Semester 5

AI Unit 1 – Introduction to Artificial Intelligence

Definition of AI, history, applications, types of agents, and basic problem-solving approaches.

1. What is Artificial Intelligence?

Artificial Intelligence (AI) is the branch of computer science that aims to build systems that can perform tasks which normally require human intelligence such as learning, reasoning, problem solving, perception and language understanding. AI systems are designed to simulate human cognitive functions and can adapt, learn from experience, and make decisions based on data and patterns.

The field of AI encompasses a wide range of techniques and approaches, from rule-based expert systems to modern machine learning and deep learning algorithms. AI systems can process vast amounts of data, recognize patterns, make predictions, and even create new content. The ultimate goal of AI is to create machines that can think, learn, and act intelligently, potentially surpassing human capabilities in specific domains.

AI is not just about creating robots or chatbots. It's about developing systems that can understand context, make informed decisions, and improve their performance over time. From recommendation systems that suggest movies to autonomous vehicles that navigate complex traffic, AI is transforming how we interact with technology and solve real-world problems.

Key Definitions

  • Thinking Humanly: Systems that try to mimic human thought process. This approach involves understanding how humans think and replicating that process computationally. Cognitive science and psychology play important roles here.
  • Acting Humanly: Systems that behave like humans (Turing test view). If a machine can fool a human into thinking it's human through conversation, it passes the Turing test. This focuses on behavior rather than internal processes.
  • Thinking Rationally: Systems that reason using logic. This approach uses formal logic and mathematical reasoning to solve problems. It's about correct reasoning, not necessarily human-like reasoning.
  • Acting Rationally: Systems that choose actions to achieve the best outcome. This is the most practical approach, focusing on optimal behavior rather than human-like behavior. Rational agents maximize expected utility.

2. Brief History of AI

The history of AI is a fascinating journey that spans over seven decades. Understanding this history helps us appreciate how AI has evolved and where it might be heading.

  • 1950 – Alan Turing proposes the Turing test in his paper "Computing Machinery and Intelligence," establishing a benchmark for machine intelligence.
  • 1956 – The Dartmouth conference marks the birth of AI as a field. John McCarthy coins the term "Artificial Intelligence." Early optimism about solving AI problems quickly emerges.
  • 1960–70 – Early search programs and expert systems are developed. Programs like ELIZA (natural language processing) and SHRDLU (natural language understanding) demonstrate early AI capabilities. The first AI winter begins due to unmet expectations.
  • 1980 – Knowledge-based systems and rule engines become popular. Expert systems like MYCIN (medical diagnosis) and DENDRAL (chemical analysis) show practical applications. Commercial AI industry emerges.
  • 1990s – Machine learning gains prominence. Support vector machines, neural networks, and statistical methods become more sophisticated. IBM's Deep Blue defeats world chess champion Garry Kasparov in 1997.
  • 2000+ – Machine Learning, Deep Learning, Big Data, and modern AI applications revolutionize the field. Breakthroughs in deep learning, availability of large datasets, and powerful computing enable unprecedented AI capabilities. Applications in image recognition, natural language processing, and autonomous systems become mainstream.

3. Applications of AI

AI applications have permeated almost every aspect of modern life. From the moment we wake up and check our smartphones to the recommendations we receive while shopping online, AI is working behind the scenes to enhance our experiences and solve complex problems.

  • Natural Language Processing – Chatbots, translators, sentiment analysis, text summarization, and voice assistants like Siri and Alexa. NLP enables machines to understand, interpret, and generate human language in valuable ways.
  • Computer Vision – Face detection, medical imaging, autonomous vehicles, quality control in manufacturing, and augmented reality. Computer vision allows machines to interpret and understand visual information from the world.
  • Robotics – Industrial robots for manufacturing, service robots for healthcare and hospitality, autonomous drones, and robotic process automation. Robotics combines AI with mechanical engineering to create intelligent machines.
  • Expert Systems – Medical diagnosis systems, credit scoring, fraud detection, and recommendation engines. Expert systems capture and apply knowledge from human experts to solve specific problems.
  • Games – Chess engines, Go programs (AlphaGo), strategy games, and game AI for video games. Game AI demonstrates problem-solving capabilities and has led to significant advances in search algorithms.
  • Healthcare – Disease diagnosis, drug discovery, personalized treatment plans, and medical image analysis. AI is revolutionizing healthcare by enabling early detection and personalized medicine.
  • Finance – Algorithmic trading, risk assessment, fraud detection, and credit approval systems. AI helps financial institutions make better decisions and manage risks more effectively.

4. Intelligent Agents

An agent is anything that perceives its environment through sensors and acts upon that environment using actuators. Agents are the fundamental building blocks of AI systems. They can range from simple programs that respond to specific inputs to complex systems that learn and adapt over time.

The concept of an agent provides a unified framework for understanding AI systems. Every AI system can be viewed as an agent that receives percepts (inputs) from its environment, processes this information, and takes actions to achieve its goals. The sophistication of an agent depends on how it processes information and makes decisions.

Types of Agents

  • Simple reflex agents – React only to current percept. They use condition-action rules: if certain conditions are met, take specific actions. These agents don't consider history or future consequences. Example: a thermostat that turns on heating when temperature drops below a threshold.
  • Model-based agents – Maintain some internal state of the world. They keep track of aspects of the world that aren't directly observable. This allows them to handle partially observable environments. Example: a robot that tracks its position even when GPS is temporarily unavailable.
  • Goal-based agents – Use goal information to choose actions. They consider future actions and their consequences. They search for sequences of actions that achieve their goals. Example: a navigation system that finds the best route to a destination.
  • Utility-based agents – Maximize an internal utility function. They consider not just whether a goal is achieved, but how well it's achieved. They make trade-offs between multiple goals. Example: a trading system that balances profit and risk.
  • Learning agents – Improve performance based on experience. They can adapt to new environments and tasks. Learning agents have components for learning, performance, and feedback. Example: recommendation systems that improve suggestions based on user interactions.

5. Problem-Solving by Search

Many AI problems can be modelled as search in a state space. Search is a fundamental technique in AI for finding solutions to problems. The idea is to explore possible states and transitions to find a path from an initial state to a goal state.

In search problems, we define a state space that represents all possible configurations of the problem. Each state represents a particular situation, and actions transform one state into another. The goal is to find a sequence of actions (a path) that leads from the initial state to a goal state.

Components of Search Problems

  • Initial state – The starting configuration of the problem. This is where the search begins.
  • Actions – The set of possible actions that can be taken in each state. Actions define how we can move from one state to another.
  • Transition model – Describes what each action does. It specifies the resulting state when an action is applied to a state.
  • Goal test – Determines whether a given state is a goal state. This tells us when we've solved the problem.
  • Path cost – A function that assigns a cost to each path. This allows us to find optimal solutions (lowest cost paths).

Examples of search problems include: 8-puzzle (sliding tile puzzle), route finding (navigation), chess search (game playing), scheduling problems, and resource allocation. Search algorithms like breadth-first search, depth-first search, A* search, and heuristic search are fundamental tools in AI problem-solving.

6. Summary

In Unit 1, focus on definitions of AI, the four views (thinking/acting humanly and rationally), timeline of AI history, major applications, and structure & types of intelligent agents. Understanding these foundational concepts is crucial for advanced AI topics.

AI is a rapidly evolving field with applications in virtually every domain. From healthcare to finance, from entertainment to transportation, AI is transforming how we live and work. As we continue to develop more sophisticated AI systems, understanding the fundamentals becomes increasingly important.

7. Frequently Asked Questions

Q1: What is the difference between AI, Machine Learning, and Deep Learning?

AI is the broad field of creating intelligent machines. Machine Learning is a subset of AI that enables systems to learn from data without explicit programming. Deep Learning is a subset of Machine Learning that uses neural networks with multiple layers to learn complex patterns.

Q2: Can AI systems think like humans?

Current AI systems don't think like humans. They process information differently, using statistical patterns and mathematical operations. While they can perform specific tasks very well, they lack general intelligence and human-like understanding.

Q3: What are the main challenges in AI?

Key challenges include: achieving general intelligence (not just narrow tasks), ensuring AI systems are safe and reliable, addressing ethical concerns, handling uncertainty and incomplete information, and creating systems that can learn from limited data.

Q4: How do intelligent agents differ from simple programs?

Intelligent agents are autonomous, perceive their environment, take actions to achieve goals, and can adapt or learn. Simple programs follow fixed instructions without autonomy or adaptation capabilities.

Q5: What is the importance of search in AI?

Search is fundamental to AI problem-solving. Many AI problems can be formulated as search problems, and efficient search algorithms enable AI systems to find solutions in complex state spaces. Search techniques are used in planning, game playing, optimization, and many other AI applications.