Unit 1: Introduction to AI
Artificial Intelligence notes · PTU syllabus (UGCC2521)
On this page
Unit summary
Artificial Intelligence (AI) is about building machines that act intelligently. This unit defines AI, introduces intelligent agents and their environments, explains rationality and agent structures, and studies knowledge-based agents (with the Wumpus World) and problem-solving agents.
After this unit you can
- Define AI and its approaches
- Describe agents, environments and the PEAS framework
- Explain rationality and the types of agent structure
- Describe knowledge-based and problem-solving agents
PTU syllabus topics
- What AI is
- intelligent agents and environments
- the concept of rationality
- agent structure
- knowledge-based agents
- the Wumpus World example
- problem-solving agents
Topic 1
What is AI?
Artificial Intelligence is the branch of computer science that builds systems able to perform tasks that normally need human intelligence — reasoning, learning, perception, language and decision-making.
Thinking humanly
Cognitive modelling
Thinking rationally
Laws of thought, logic
Acting humanly
Turing test
Acting rationally
Rational agents (the modern approach)
The Turing Test (1950): a machine is intelligent if a human interrogator cannot tell it apart from a human in conversation.
Topic 2
Agents and environments
A rational agent chooses the action expected to maximise its performance measure given its percept history
An agent perceives its environment through sensors and acts on it through actuators. Its behaviour is defined by an agent function mapping percept sequences to actions. PEAS describes a task: Performance measure, Environment, Actuators, Sensors.
Example
Self-driving car — P: safety, speed, comfort; E: roads, traffic, pedestrians; A: steering, brakes, accelerator; S: cameras, GPS, LIDAR, speedometer.
| Environment property | Opposite |
|---|---|
| Fully observable | Partially observable |
| Deterministic | Stochastic |
| Episodic | Sequential |
| Static | Dynamic |
| Discrete | Continuous |
| Single-agent | Multi-agent |
Example
Classifying two environments: chess with a clock — fully observable, deterministic (strategic), sequential, semi-dynamic, discrete and multi-agent. Taxi driving — partially observable, stochastic, sequential, dynamic, continuous and multi-agent. Taxi driving is harder on every property, which is why self-driving cars need far more sensing and learning than a chess program.
Exam tip
For PEAS questions, write the four headings as a table and give at least three items under each — performance measure, environment, actuators and sensors.
Topic 3
Rationality and agent structure
A rational agent chooses the action expected to maximise its performance measure, given its percepts and knowledge. Rational ≠ perfect — it does the best with the information it has.
Simple reflex
Acts on the current percept with if-then rules
Model-based reflex
Keeps an internal model of the world
Goal-based
Chooses actions that reach a goal
Utility-based
Chooses actions that maximise happiness (utility)
Learning agent
Improves its performance from experience
Example
One example of each agent type: a room thermostat that switches on below 20°C is a simple reflex agent; a robot vacuum that remembers which areas it has cleaned is model-based; a GPS app that plans a route to a destination is goal-based; a ride-hailing app that picks the trip balancing time, fare and driver rating is utility-based; a spam filter that improves as users mark emails is a learning agent.
Topic 4
Knowledge-based agents and the Wumpus World
A knowledge-based agent has a knowledge base (KB) of facts in a formal language and an inference engine that derives new facts. It TELLs the KB what it perceives and ASKs it what to do. Wumpus World: a 4 × 4 cave with a Wumpus (monster), pits and gold. The agent senses stench (Wumpus nearby), breeze (pit nearby) and glitter (gold here), and uses logical reasoning to move safely, grab the gold and climb out.
Topic 5
Problem-solving agents
A problem-solving agent decides what to do by searching for a sequence of actions that reaches a goal. A problem is defined by: initial state, actions, transition model, goal test and path cost.
Example
8-puzzle — states: tile arrangements; actions: move the blank up, down, left or right; goal: tiles in order; path cost: number of moves.
Key terms
- Agent
- An entity that perceives and acts in an environment
- PEAS
- Performance, Environment, Actuators, Sensors
- Rational agent
- An agent that acts to maximise expected performance
- Knowledge base
- A set of facts and rules in a formal language
- Turing Test
- A test of machine intelligence through conversation
Quick revision
- Modern AI approach: acting rationally.
- Agent = sensors + actuators + agent function.
- Agents: simple reflex, model-based, goal-based, utility-based, learning.
- Problem = initial state, actions, transition, goal test, path cost.
Important exam questions
Practice questions written to the PTU exam pattern for this unit's syllabus: short answers (Section A style) and long answers (Sections B and C style).
Short-answer questions
- Q1.Define artificial intelligence.
- Q2.What is the Turing Test?
- Q3.Define an agent and give PEAS for a vacuum-cleaner agent.
- Q4.What is a rational agent?
- Q5.Differentiate between goal-based and utility-based agents.
Long-answer questions
- Q1.Explain the types of agents with diagrams.
- Q2.Explain the properties of task environments with examples.
- Q3.Describe the Wumpus World and how a knowledge-based agent reasons in it.
- Q4.Explain how a problem is formulated for a problem-solving agent, with an example.
Stuck on this unit?
Message SBS on WhatsApp for help with Artificial Intelligence, or to ask about studying BCA at Synetic.
