Artificial Intelligence

Artificial intelligence, or AI, allows computer systems to perform tasks that usually require human intelligence. These tasks may include recognising images, understanding language, finding patterns, making predictions and creating content.

AI is already present in smartphones, search engines, online shopping, banking, healthcare and entertainment.

Think of AI as a very fast assistant. It can process a huge amount of information, but it can still misunderstand instructions or confidently produce a wrong answer.

Simple rule: Use AI for assistance, not unquestioned authority.

What Is Artificial Intelligence?

Artificial intelligence refers to machine-based systems that use input data to generate outputs such as predictions, recommendations, decisions or content.

The OECD definition of an AI system also recognises that AI systems can operate with different levels of autonomy and may influence physical or digital environments.

AI systems may:

Recognise a face
Translate a language
Recommend a film
Detect suspicious payments
Answer questions
Generate an image
Control a robot
Predict equipment failure

How Does AI Work?

A basic AI system follows this process:

Data → Training → AI Model → Input → Output

1. Data

The system receives examples or information.

For example, an image-recognition system may receive thousands of labelled images of cats and dogs.

2. Training

An algorithm searches for patterns in the data.

It may learn that cats and dogs have different:

Ear shapes
Facial structures
Body sizes
Fur patterns

3. AI Model

The patterns learned during training are stored in a model.

4. New Input

The trained model receives an image it has not seen before.

5. Output

The model predicts whether the image contains a cat or dog.

The answer is a prediction based on learned patterns. It is not human understanding.

A Simple Real-Life Example

Imagine teaching a child to recognise apples.

You show the child many examples:

Red apples
Green apples
Large apples
Small apples

Eventually, the child identifies a new apple.

Machine learning follows a somewhat similar idea, but it uses data, mathematics and algorithms rather than human experience.

If the training data contains only red apples, the system may struggle to recognise a green one. This shows why diverse and accurate data matters.

AI, Machine Learning and Deep Learning

These terms are connected but not identical.

AI is the broad field of creating computer systems that perform intelligent tasks.

Machine Learning

Machine learning, or ML, is a method that allows computers to learn patterns from data.

Deep Learning

Deep learning is a type of machine learning that uses multi-layered artificial neural networks.

Generative AI

Generative AI creates new content such as text, images, audio, video or code.

The relationship can be understood as:

Artificial Intelligence → Machine Learning → Deep Learning → Many Generative AI Systems

Not every AI system uses machine learning, and not every machine-learning system generates content.

Artificial Intelligence vs Traditional Programming

In traditional programming, a developer writes rules for the computer to follow.

Rules + Data → Answer

In machine learning, the computer learns patterns from examples.

Data + Expected Answers → Learned Model

Traditional example

A programmer creates a rule:

If the temperature is above 30°C, display “Hot.”

Machine-learning example

A system studies years of weather data to predict tomorrow’s temperature.

Traditional Programming	Machine Learning
Rules are written directly	Patterns are learned from data
Suitable for clear instructions	Suitable for complex patterns
Easier to explain	May be harder to explain
Example: Tax calculation	Example: Fraud detection

Main Types of Artificial Intelligence

Narrow AI

Narrow AI is designed for a particular task or limited set of tasks.

Examples:

Spam filtering
Voice recognition
Navigation
Product recommendations
Image classification
AI chat assistants

Most practical AI systems today are forms of narrow AI.

Artificial General Intelligence

Artificial general intelligence, or AGI, usually refers to a hypothetical system able to perform a broad range of intellectual tasks at a human-like level.

AGI is a research goal and debated concept. It should not be confused with today’s specialised AI systems.

Artificial Superintelligence

Artificial superintelligence is a hypothetical idea of AI exceeding human intelligence across nearly every area.

It currently belongs to future discussion rather than established everyday technology.

Main Types of Machine Learning

Supervised Learning

Supervised learning uses labelled examples.

For example, emails may be labelled:

Spam
Not spam

The system learns patterns from the labelled emails and predicts the category of a new message.

Common uses include:

Image classification
Fraud detection
Price prediction
Medical-image assistance

Unsupervised Learning

Unsupervised learning receives data without labelled answers and searches for patterns or groups.

For example, a shop may use it to group customers according to purchasing behaviour.

Common uses include:

Customer grouping
Pattern discovery
Unusual-activity detection
Data exploration

Reinforcement Learning

Reinforcement learning learns through actions, rewards and penalties.

For example, a game-playing system may receive:

A reward for winning
A penalty for losing

Over many attempts, it learns which actions produce better results.

Possible uses include:

Robotics
Game-playing systems
Resource management
Route planning

What Is Deep Learning?

Deep learning uses artificial neural networks containing several processing layers.

It is especially useful for:

Image recognition
Speech recognition
Language processing
Video analysis
Generative AI

Deep-learning systems often require large amounts of data and computing power.

The word “neural” is inspired by biological brains, but artificial neural networks are mathematical systems and do not work exactly like human brains.

What Is Generative AI?

Generative AI creates new content based on patterns learned during training.

It may generate:

Text
Images
Music
Speech
Video
Computer code
Presentations

Simple example

A user asks:

“Write a beginner-friendly explanation of computer memory.”

The AI generates a new response based on patterns learned from large amounts of text.

It does not simply search for and copy one stored answer.

Large Language Models

A large language model, or LLM, is trained on large amounts of text and learns statistical relationships between words and pieces of language.

LLMs can:

Answer questions
Summarise text
Translate languages
Generate code
Rewrite documents
Explain topics
Analyse supplied text

An LLM predicts suitable language based on patterns. It may produce convincing information that is incomplete or incorrect.

What Are AI Hallucinations?

An AI hallucination occurs when a generative AI system produces false or unsupported information as though it were correct.

Examples include:

Inventing a book
Giving a fake web link
Creating an incorrect quotation
Producing a wrong calculation
Confusing two people
Describing an event that did not happen

Important AI-generated claims should be checked using reliable sources.

Think of AI like an enthusiastic assistant: fast, helpful and occasionally certain about something it completely invented.

What Are AI Agents?

An AI agent is a system designed to perform actions toward a goal, sometimes using applications, tools or external information.

For example, an AI agent might:

Read a meeting request.
Check a calendar.
Compare available times.
Draft a response.
Create a reminder after receiving permission.

Agents can automate multi-step tasks, but they require:

Clear permissions
Human approval
Security controls
Activity logs
Spending limits
Error handling

An agent should not receive unlimited access simply because it is convenient.

Major Areas of AI

Natural Language Processing

Natural language processing helps computers work with human language.

Examples:

Translation
Chatbots
Sentiment analysis
Text summarisation
Speech-to-text

Computer Vision

Computer vision helps machines analyse images and videos.

Examples:

Medical-image analysis
Facial recognition
Quality inspection
Object detection
Security-camera analysis

Speech Recognition

Speech-recognition systems convert spoken language into text or commands.

Examples:

Voice typing
Digital assistants
Automated subtitles
Call transcription

Robotics

AI can help robots:

Recognise objects
Plan movement
Avoid obstacles
Adapt to changing conditions
Perform physical tasks

Expert Systems

Expert systems use stored knowledge and rules to assist with decisions in a specialised area.

Recommendation Systems

Recommendation systems predict what a user may want to watch, read or buy.

Real-Life Uses of Artificial Intelligence

Healthcare

AI may assist with:

Medical-image analysis
Patient-risk prediction
Drug research
Appointment management
Health monitoring

AI should support qualified medical professionals rather than independently replace clinical judgement.

Education

AI can help:

Explain difficult concepts
Generate practice questions
Provide language assistance
Support accessibility
Give personalised feedback

Students should still check facts and follow their institution’s rules on AI use.

Banking and Finance

AI is used for:

Fraud detection
Credit-risk analysis
Customer support
Document processing
Market analysis

High-impact financial decisions require fairness, transparency and appropriate human oversight.

Cybersecurity

AI can help detect:

Suspicious logins
Malware patterns
Phishing messages
Unusual network activity
Fraudulent behaviour

Attackers may also use AI to create convincing scams, automate attacks or generate malicious content.

Transport

AI may support:

Route planning
Traffic prediction
Driver assistance
Delivery systems
Vehicle maintenance

Entertainment

AI is used for:

Film recommendations
Music recommendations
Game characters
Content creation
Automatic subtitles

Agriculture

AI systems can analyse:

Crop images
Soil data
Weather patterns
Animal health
Irrigation needs

Business

Businesses may use AI for:

Customer service
Sales forecasting
Inventory planning
Document processing
Quality checks
Data analysis

Benefits of Artificial Intelligence

AI can:

Process large amounts of data
Automate repetitive tasks
Work continuously
Identify hidden patterns
Improve accessibility
Support faster decisions
Personalise services
Assist scientific research
Reduce dangerous manual work

The value of AI depends on how accurately and responsibly it is designed and used.

Limitations of Artificial Intelligence

AI has important limitations.

AI Can Be Wrong

AI outputs are predictions, not guaranteed facts.

AI Depends on Data

Incorrect, incomplete or biased data may produce poor results.

AI May Lack Context

A model may misunderstand humour, culture, personal circumstances or unusual situations.

AI May Be Difficult to Explain

Some complex models cannot easily explain why they produced a particular decision.

AI Does Not Have Human Experience

AI does not experience emotions, responsibility or consequences like a human.

AI Can Be Manipulated

Attackers may attempt to change inputs, training data or instructions to influence the system.

AI Requires Resources

Large AI systems may require significant computing power, electricity and cooling.

Bias in Artificial Intelligence

AI bias occurs when a system produces unfair or systematically different outcomes.

Possible causes include:

Unbalanced training data
Historical discrimination
Incorrect labels
Poor system design
Inappropriate use
Lack of testing

For example, an image system trained mostly on one group of people may perform less accurately for others.

Reducing bias requires:

Diverse data
Regular testing
Human review
Clear documentation
Monitoring after deployment
A way to challenge important decisions

AI Privacy Risks

AI systems may process:

Personal messages
Photographs
Voice recordings
Location data
Health information
Work documents
Biometric data

Before using an AI service, ask:

What information am I sharing?
Is it confidential?
Who can access it?
How long is it stored?
Is it used for training?
Can I delete it?
Does my school or employer allow it?

Do not paste passwords, bank information or confidential documents into an AI tool without proper authorisation and protection.

AI Security and Adversarial Machine Learning

AI systems can become targets of specialised attacks.

Examples include:

Data Poisoning

An attacker introduces harmful or misleading information into training data.

Evasion Attack

An attacker modifies an input to confuse an AI model.

Prompt Injection

A malicious instruction attempts to make a generative AI system ignore its intended rules.

Model Theft

An attacker attempts to copy or extract information about an AI model.

Privacy Attack

An attacker attempts to discover sensitive information used by the system.

NIST’s Adversarial Machine Learning guidance provides a technical taxonomy for attacks and mitigations across the AI life cycle.

Deepfakes

Deepfakes are realistic-looking synthetic images, audio or videos created or changed using AI.

They may be used for:

Entertainment
Accessibility
Education
Film production

They may also be misused for:

Fraud
Fake evidence
Impersonation
Political manipulation
Harassment
False information

Do not trust a video or voice message only because it looks or sounds realistic. Verify unusual requests through another trusted method.

Responsible and Ethical AI

Responsible AI aims to make systems:

Safe
Fair
Transparent
Secure
Privacy-protecting
Accountable
Reliable
Human-centred

UNESCO’s Recommendation on the Ethics of Artificial Intelligence highlights principles including safety, privacy, fairness, non-discrimination and human rights.

The NIST AI Risk Management Framework organises AI risk management around four functions:

Govern
Map
Measure
Manage

Responsible AI is not a single setting. It requires continuous testing and monitoring throughout the system’s life cycle.

How to Use AI Safely

1. Protect Sensitive Information

Do not share confidential data without permission.

2. Verify Important Answers

Check important claims using reliable sources.

3. Understand the Purpose

Use the tool only for tasks it was designed to support.

4. Keep Human Oversight

People should review decisions that affect health, employment, education, finance or safety.

5. Review Permissions

Do not give an AI agent unnecessary access to email, files, payments or accounts.

6. Check for Bias

Test whether the system works fairly for different users and situations.

7. Record AI Use

Organisations should document where AI is used and who is responsible.

8. Follow Rules

Students and employees should follow their institution’s policies on AI-generated work.

AI and Jobs

AI can automate some tasks, change existing roles and create new types of work.

Tasks likely to change may include:

Routine data entry
Basic document processing
Simple customer queries
First-draft content creation
Repetitive analysis

Human skills remain important for:

Judgement
Responsibility
Communication
Creativity
Leadership
Empathy
Ethical decisions
Complex problem-solving

Learning to work with AI may become more useful than trying to compete with it at repetitive tasks.

Skills for Learning Artificial Intelligence

Beginners can start with:

Computer fundamentals
Basic mathematics
Statistics
Python programming
Data handling
Machine learning basics
Cybersecurity
Critical thinking
Communication
AI ethics

You do not need to build a huge AI model to understand AI. Start with small datasets and simple projects.

Common AI Myths

Myth: AI is always correct

Fact: AI can produce incorrect or invented information.

Myth: AI thinks exactly like a human

Fact: AI processes patterns using algorithms and data. It does not have ordinary human experience.

Myth: Every automated program is AI

Fact: A basic calculator or fixed rule-based script may not be considered AI.

Myth: AI will immediately replace every job

Fact: AI affects tasks and roles differently. Many jobs will change rather than disappear completely.

Myth: More data always creates better AI

Fact: Data quality, relevance, diversity and legality matter as much as quantity.

Myth: AI is completely objective

Fact: AI can reflect bias from its data, design and use.

A Fun AI Example

Imagine an AI system trained to identify healthy fruit.

During training, every healthy apple is photographed on a white plate. Every damaged apple is photographed on a dark table.

The system may learn:

White background = healthy
Dark background = damaged

It appears accurate during testing but has learned the wrong pattern.

This is why AI developers must carefully examine:

Training data
Testing methods
Real-world conditions
Unexpected behaviour

Explore Artificial Intelligence Topics

This guide connects to the following detailed lessons:

Machine Learning
Deep Learning
Generative AI
Large Language Models
Natural Language Processing
Computer Vision
AI Agents
AI in Cybersecurity
AI Ethics and Bias
Deepfakes
Future of Artificial Intelligence

Conclusion

Artificial intelligence enables computer systems to generate content, recognise patterns, make predictions and support decisions. Machine learning, deep learning and generative AI are important parts of the field.

AI can improve healthcare, education, cybersecurity and accessibility. However, it can also produce errors, bias, privacy risks and security problems.

Remember: AI can be powerful without being perfect, intelligent without being human and useful without being trusted blindly.

Frequently Asked Questions

What is artificial intelligence?

Artificial intelligence is the field of creating computer systems that can perform tasks such as learning patterns, understanding language, recognising images and making predictions.

What is the difference between AI and machine learning?

AI is the wider field. Machine learning is one method used to create AI systems by learning patterns from data.

What is generative AI?

Generative AI creates new content such as text, images, audio, video or computer code.

Can artificial intelligence make mistakes?

Yes. AI can misunderstand instructions, reflect bias and generate incorrect or invented information.

Is AI the same as a robot?

No. AI is software or a computing capability. A robot is a physical machine that may or may not use AI.

Is AI safe?

AI can be useful and safe when properly designed, tested, secured and monitored. Its risks depend on the system and how it is used.