📚 Knowledge Library — Topic 6.4 — Automated & Emerging Technologies

Artificial Intelligence Explained Simply

Understand what makes a system intelligent, how machine learning improves through experience, and how expert systems reach decisions.

Start LearningWorked Example
1. Invitation

Most programs follow rules. AI can use rules to make intelligent decisions.

Artificial intelligence is the simulation of intelligent behaviour by computers.

An AI system may analyse data, recognise patterns, reason, make decisions, learn and adapt.

💡 Key idea: AI uses data and rules to perform tasks that normally require human intelligence.
Figure 1.1
AI at Work
Collect data
↓
Find patterns
↓
Make decisions
2. Big Idea

AI characteristics describe what the system can do.

An AI system may be able to reason, learn, adapt, recognise patterns and make decisions.

These are characteristics. Machine learning and expert systems are types of AI.

🎯 Exam Tip: characteristics describe abilities. They are not examples or applications of AI.
Figure 2.1
Characteristics
Reason
Learn
Adapt
Decide
3. FutureLogic Bridge

Think of a chess player who improves after every game.

The player remembers successful moves and avoids moves that led to defeat.

Over time, the player becomes better because experience changes future decisions.

💡 Bridge: machine learning is practice that changes the next decision.
Figure 3.1
Learning from Experience
Try
↓
Check result
↓
Improve
4. Machine Learning

Machine learning allows a system to adapt its own processes.

The system collects data, identifies patterns and records successful or unsuccessful outcomes.

It then changes its rules, data or processes so that future results become more accurate or useful.

💡 Machine learning = learn from results and adapt future behaviour.
Figure 4.1
Learning Cycle
Data
↓
Pattern
↓
Adjustment
↺
5. Supervised and Unsupervised

Some systems are taught; others search for patterns themselves.

In supervised learning, the system is given examples with the correct answers.

In unsupervised learning, the system receives data and finds groups or patterns without being told the answer.

💡 Supervised = labelled examples. Unsupervised = discover the pattern.
Figure 5.1
Two Approaches
Examples + answers
or
Data only
6. Expert Systems

An expert system gives advice in one specialist area.

An expert system attempts to reproduce the knowledge and decision-making of a human expert.

It may be used for medical diagnosis, car fault finding, financial advice or technical support.

💡 An expert system knows one field well; it is not a general-purpose thinker.
Figure 6.1
Specialist Advice
User answers
↓
System applies rules
↓
Advice or diagnosis
7. Four Components

Every expert system has four main parts.

ComponentPurpose
InterfaceAllows the user to enter answers and receive results.
Knowledge baseStores facts and specialist knowledge.
Rule baseStores the rules used to connect and apply the facts.
Inference engineApplies the rules, decides questions and reaches the result.
Figure 7.1
Expert System Parts
Interface
Knowledge base
Rule base
Inference engine
8. The Inference Engine

The inference engine appears twice in the process.

First, it uses previous answers to decide which question should be asked next.

Later, it applies the rules to the stored knowledge to reach the diagnosis or recommendation.

🎯 Exam Tip: inference engine = decides the next question and decides the final result.
Figure 8.1
Two Decisions
What to ask?
↓
What is the result?
9. Worked Example

A search engine learns which results are useful.

The search engine records which results users select and identifies patterns in their searches.

Exam-style process

It stores successful and unsuccessful results, changes its rules or ranking data and uses the new information when a similar search is made.

Model answer: “The search engine collects data about previous searches and identifies patterns in the results selected. It changes its rules or data so that more relevant results are shown for similar searches in future.”
Figure 9.1
Improving Search Results
Previous searches
↓
Identify patterns
↓
Better future results
10. Common Mistake

Do not give a generic definition when the question gives a context.

“The system learns and adapts” is too vague for a question about a game, robot or search engine.

⚠️ Better answer: explain what data the system collects, what pattern it identifies and exactly what future action it improves.
Figure 10.1
Apply the Context
Learns from what?
↓
Improves what?
11. Another Common Mistake

An expert system needs a user to provide information.

An expert system asks questions through its interface and uses the answers to reach a result.

It is therefore unsuitable for an autonomous robot that must make decisions without a user answering questions.

⚠️ Common Mistake: suggesting an expert system whenever a question mentions AI.
Figure 11.1
User Required
Question
↓
User answer
↓
Next decision
12. Summary

Artificial intelligence in one screen.

Artificial intelligence simulates intelligent behaviour using data and rules.

Machine learning allows a system to learn from results and adapt its own rules, data or processes.

An expert system uses an interface, knowledge base, rule base and inference engine to provide specialist advice.

💡 Final thought: AI analyses and decides; machine learning improves; expert systems advise.
Figure 12.1
Final Model
AI
↓
Machine learning
+
Expert systems