Training Day → Hour 1 → AI Basics

🧠 AI Basics

Before we talk about ChatGPT and Copilot, we need to understand what happens behind the scenes — how computers learn and how they make decisions.

⏱ 35 Minutes 📖 Theoretical Lesson ✏️ Closing Exercise

🤖 What is Artificial Intelligence?

Artificial Intelligence (AI) is a field of computer science that aims to build software capable of mimicking human abilities — such as understanding, learning, problem-solving, and decision-making.

💡
Think of it this way

Humans learn from experience — a child touches a stove once and then avoids it forever. AI learns from data — the more examples it sees, the smarter it gets.

AI is not a single thing — it is an umbrella under which overlapping fields exist. Look at how these three terms overlap:

Artificial Intelligence Artificial Intelligence
Software that mimics human capabilities in general
Machine Learning Machine Learning
The model learns from data instead of manually written rules
Deep Learning Deep Learning
Multi-layered neural networks to extract complex patterns
🔑
The Essential Element: Data

Without data, there is no learning. Every AI model you see today — ChatGPT, Copilot, Google Translate — has been trained on massive amounts of text, images, and sounds generated by humans.

📚 Types of Learning

How does a computer learn? There are three different methods — each method suits a specific type of problem.

🏷️

Supervised Learning

Supervised Learning

The model learns from pre-labeled examples — each example comes with the correct answer. After training, it can predict answers for new examples it has never seen before.

Real-world Example Thousands of cat and dog images with their labels → the model learns the difference → it classifies new images on its own
🔍

Unsupervised Learning

Unsupervised Learning

No ready-made answers — the model searches for hidden patterns and similarities within the data on its own and discovers groups without anyone telling it what to look for.

Real-world Example Unclassified customer data → the model discovers: these are young people who prefer delivery, and these are families who care about price
🎮

Reinforcement Learning

Reinforcement Learning

The model learns by trial and error — it tries, gets a reward if it succeeds or a penalty if it fails, and repeats until it masters the task.

Real-world Example A program learns to play chess → tries moves → wins or loses → gradually becomes an expert

🎯 What Does the Model Do?

When the model learns, it is employed in one of these three main tasks:

📂

Classification

Classification

Determining which category a specific item belongs to from pre-defined categories.

Is this email normal or spam?
🫧

Clustering

Clustering

Grouping similar data together without prior knowledge of the categories.

What are the natural groups in this data?
📈

Regression

Regression

Predicting a continuous numerical value based on given features.

How much will this house cost?

🔤 Concepts to Complete the Picture

🧪
Turing Test
Turing Test

Proposed by mathematician Alan Turing in 1950. The idea: if you talk to a computer and cannot distinguish it from a human, it passes the test.

🗄️
Big Data
Big Data

Massive amounts of data that exceed the processing capability of traditional tools. It requires special techniques to store, analyze, and extract value from it.

🕸️
Deep Learning
Deep Learning

A branch of machine learning that uses multi-layered artificial neural networks. It is what power image and speech recognition, and language translation.

📦
Model
Model

The final output of the training process — a mathematical or software representation that stores what it learned from data and uses it to predict or classify new data.

✏️ Exercise: Do you remember?

Click on each term to check its definition — was your answer correct?

🧠 Terminology Review Revealed 0 / 10
Supervised Learning →
The model learns from data attached to correct answers, such as labeled images: cat / dog.
Unsupervised Learning →
The model learns from unlabeled data and discovers patterns or hidden groups on its own.
Reinforcement Learning →
The model learns through trial and error, receiving rewards or penalties after each decision.
Classification →
Determining the category an item belongs to — example: normal email or spam?
Clustering →
Grouping similar data together without prior knowledge of the categories.
Regression →
Predicting a continuous numerical value — such as predicting a house price based on its specifications.
Turing Test →
A test to determine if a computer can mimic human intelligence to the point of fooling a human.
Big Data →
Massive amounts of diverse data that require special techniques to process and analyze.
Deep Learning →
A branch of machine learning that uses multi-layered neural networks to extract complex patterns.
Model →
A mathematical or software representation that learns from data to perform a task such as prediction or classification.