Introduction to Artificial Intelligence
What is Artificial Intelligence?
Artificial Intelligence (AI) is a branch of computer science focused on building computer systems that can perform tasks that usually require human intelligence. But what does that actually mean?
When we say "human intelligence," we mean abilities like:
- Learning: The ability to gain new knowledge from experience
- Reasoning: The ability to draw conclusions from information
- Problem-solving: Finding solutions in new situations
- Language understanding: Understanding and producing natural language
- Perception: Recognizing and understanding images, sounds, and the environment
- Planning: Building plans to achieve goals
Artificial intelligence tries to create computer systems that can perform some of these abilities— or even all of them.
In regular programming, the programmer writes exactly what the computer should do in every situation. In AI, the computer learns by itself what to do based on examples and experience.
How does AI work? The basic idea
At its core, most of today's AI systems work on a simple principle: learning from examples.
Example: Teaching a computer to recognize cats
Suppose we want to teach a computer to recognize cats in images:
- Step 1 - Data: Show the computer millions of images, some with cats and some without
- Step 2 - Training: The computer analyzes the images and looks for patterns common to cat images (pointed ears, certain eyes, face shape...)
- Step 3 - Model: At the end of the process, the computer has "built" an internal model that represents "what a cat is"
- Step 4 - Use: Now when we show the computer a new image, it can check whether it matches its model
This is fundamentally different from regular programming. A programmer cannot write explicit rules for recognizing cats— there are too many variations. But AI can learn it by itself from examples.
Subfields of Artificial Intelligence
AI is a broad field that includes many subfields:
🧠 Machine Learning (ML)
The most popular branch. Computers that learn from data and improve over time. Almost all the AI we see today is based on machine learning.
Examples: Netflix recommendations, spam detection, weather forecasting
🔮 Deep Learning (DL)
A subfield of machine learning that uses artificial neural networks— structures that mimic (in a simplified way) the human brain. This is the technology behind most of the impressive progress in recent years.
Examples: ChatGPT, face recognition, automatic translation
💬 Natural Language Processing (NLP)
The ability of computers to understand, interpret, and produce human language. This is what lets us talk to Siri or chat with ChatGPT.
Examples: Chatbots, translation, text summarization, sentiment analysis
👁️ Computer Vision (CV)
The ability of computers to "see" and understand images and video. This includes object detection, faces, motion, and more.
Examples: Face ID, autonomous vehicles, image filtering
🤖 Robotics
Combining AI with physical machines. The robot must perceive the world, make decisions, and perform physical actions.
Examples: Industrial robots, robot vacuums, drones
🎮 Generative AI
The new star! Systems that can create new content— text, images, music, video, and code. This is the revolution we're experiencing now.
Examples: ChatGPT, DALL-E, Midjourney, GitHub Copilot
Types of AI by capability
AI scientists divide artificial intelligence into three levels:
1. Narrow AI (Weak AI) 🎯
A system that specializes in one specific task and excels at it. This is what we have today—and it's already very impressive!
Characteristics:
- Specializes in one task or a few related tasks
- Cannot transfer knowledge from one task to another
- Doesn't really "understand"—only recognizes patterns
- Can be excellent at its task—even better than a human
Examples:
- ChatGPT—great at conversation, but can't drive a car
- AlphaGo—beats world champions at Go, but can't play chess
- Face recognition system—recognizes faces, but doesn't understand emotions
2. General AI (AGI - Artificial General Intelligence) 🧠
A system that can perform any intellectual task a human can perform, and transfer knowledge from one task to another.
Hypothetical characteristics:
- Ability to learn anything new
- Transfer of knowledge between domains
- Real understanding, not just pattern recognition
- Ability to act in the physical world
- Perhaps: self-awareness?
Status: Does not exist. There is huge debate about when and if it will happen— predictions range from 10 years to never.
3. Superintelligence (ASI) 🚀
A system smarter than all humans combined in every domain— science, creativity, social skills, emotions.
This is a topic of speculation and science fiction. We have no idea if it's possible, when it might happen, or what it would mean. Some scientists are worried, some are optimistic, and most say it's too early to worry.
Detailed examples from everyday life
Artificial intelligence is already all around us. Here's how it works:
🎬 Netflix and YouTube - recommendation engines
What happens: Netflix knows exactly what to recommend to you
How it works:
- The system collects data: what you watched, for how long, when you stopped, what you rated
- Compares you to millions of other users with similar taste
- Analyzes the content itself: genres, actors, pace, colors
- Combines everything and calculates: "What's the chance this user will enjoy this film?"
Fact: 80% of what people watch on Netflix comes from AI recommendations!
📱 Face ID
What happens: The iPhone recognizes your face in a fraction of a second
How it works:
- Projects 30,000 infrared points onto your face
- Creates an accurate 3D map
- A neural network compares it to the stored map
- Works with glasses, beard, makeup, hat
Fact: The chance of someone else unlocking your phone: 1 in a million!
📧 Spam filtering - Gmail
What happens: Less than 0.1% of spam reaches your inbox
How it works:
- Analyzes content: suspicious words, links, images
- Checks sender: reputation, history, authentication
- Learns from your behavior: what you mark as spam
- Learns from everyone: what billions of users mark
Fact: Gmail blocks 10 million spam messages per minute!
🚗 Waze - smart navigation
What happens: Always finds the fastest route
How it works:
- Collects real-time location from millions of drivers
- Detects traffic jams, accidents, obstacles
- Learns traffic patterns: which hours have congestion on which roads
- Calculates expected time for each possible route
- Updates in real time if conditions change
Fact: Waze saves drivers an average of 5–10 minutes per trip!
💬 ChatGPT - conversation with AI
What happens: You can chat in natural language and get smart answers
How it works:
- Trained on trillions of words from the internet, books, articles
- Learned the structure of language—grammar, style, context
- Learned facts about the world from the texts
- When it receives a question—"predicts" the most likely answer word by word
Important to understand: ChatGPT doesn't "know" or "understand"—it recognizes statistical patterns. Sometimes it makes mistakes, invents things, or "hallucinates."
History of Artificial Intelligence
The story of AI is one of dreams, disappointments, and breakthroughs—from Charles Babbage and the Analytical Engine (1837), through the Turing Test (1950) and the Dartmouth Conference (1956), to ChatGPT and GPT-5. For a detailed timeline with dates, names, and developers through January 2026, go to the dedicated page:
Why it matters: impact on the world
Artificial intelligence is changing almost every area of our lives. This isn't a technology of the future—it's happening now:
🏥 Medicine
- Diagnosis: AI detects cancer in scans with 95% accuracy—sometimes better than a doctor
- Drug development: What took 10 years now takes a year with AI
- Surgery: Robots perform more precise operations
- Predictions: Detecting diseases before symptoms appear
🎓 Education
- Personalized learning: A custom curriculum for each student
- Teacher available 24/7: AI that answers questions any time
- Instant feedback: Grading assignments and giving feedback in real time
- Accessibility: Translation, read-aloud, adaptation for learning disabilities
🚗 Transportation
- Autonomous vehicles: Already driving in some cities
- Maintenance: Predicting failures before they happen
- Traffic management: Smart traffic lights that adapt
- Logistics: Optimal delivery routes
💼 Work and business
- Automation: AI performs routine tasks
- Analysis: Processing huge amounts of information
- Customer service: Chatbots 24/7
- Creativity: Help with writing, design, code
🎨 Creativity and entertainment
- Art: Creating images, music, video
- Games: Smarter opponents, adaptive worlds
- Film: Special effects, dubbing, translation
- Writing: Help creating content of all kinds
- Jobs that will disappear or change
- Questions of privacy and surveillance
- Bias and discrimination in algorithms
- Misinformation and deepfakes
- Complex ethical questions
We'll discuss all of these later, on the page about AI ethics.
Key concepts to know
📊 Algorithm
A precise set of instructions for performing a task. Like a recipe—step by step.
🧠 Neural network
A mathematical structure that mimics (in a very simplified way) the brain. Layers of "neurons" that pass information between them.
📚 Data
The fuel of AI. The more quality data there is, the better the AI learns.
🎯 Model
The result of the learning process—the "brain" that the AI built from the data.
🏋️ Training
The process in which the AI learns from data. Can take days to months.
💬 Prompt
The instruction you give to the AI. How you phrase the request greatly affects the result.
Summary 📝
- AI = computers that learn from examples instead of being programmed for every situation
- Already here: Spam, Netflix, Waze, Face ID, ChatGPT—all AI
- Types: Machine learning, deep learning, NLP, computer vision, generative AI
- Levels: We have narrow AI (specialized in one task). General AI doesn't exist yet
- History: 70 years of ups and downs, now in a revolution
- Impact: Changing medicine, education, transportation, work—everything
- Important: There are challenges too—jobs, privacy, bias, ethics
📝 Test yourself
Answer 10 questions to check your understanding. Choose one answer per question.