An overview of artificial intelligence, its history, and key concepts, including machine learning, neural networks, and natural language processing.
Introduction to Artificial Intelligence (AI)
Quick reference guide – concise, organized, and easy to read.
1. What Is AI?
| Definition |
Key Idea |
| Artificial Intelligence |
The field of computer science that creates systems capable of performing tasks that normally require human intelligence. |
| Goal |
Enable machines to perceive, reason, learn, and act autonomously or with minimal human intervention. |
2. Core Areas of AI
| Area |
Typical Tasks |
Example Applications |
| Machine Learning (ML) |
Learning patterns from data; making predictions |
Spam detection, recommendation engines |
| Natural Language Processing (NLP) |
Understanding and generating human language |
Chatbots, sentiment analysis |
| Computer Vision |
Interpreting visual information |
Facial recognition, autonomous driving |
| Robotics |
Physical interaction with the environment |
Manufacturing robots, delivery drones |
| Planning & Reasoning |
Solving problems through logical steps |
Scheduling, game playing |
| Expert Systems |
Encoding domain knowledge for decision support |
Medical diagnosis, fault detection |
3. Fundamental Concepts
3.1. Data → Information → Knowledge → Wisdom (DIKW Pyramid)
- Data – Raw facts (e.g., sensor readings).
- Information – Processed data (e.g., aggregated statistics).
- Knowledge – Patterns and relationships (e.g., a model).
- Wisdom – Insightful decisions based on knowledge.
3.2. Types of Learning
| Type |
Description |
Typical Use |
| Supervised |
Model learns from labeled examples (input‑output pairs). |
Image classification |
| Unsupervised |
Model discovers structure in unlabeled data. |
Customer segmentation |
| Semi‑supervised |
Small labeled set + large unlabeled set. |
Speech recognition |
| Reinforcement |
Agent learns via trial‑and‑error, receiving rewards/punishments. |
Game playing, robotics |
3.3. Model vs. Algorithm
- Algorithm – Procedure or set of rules (e.g., gradient descent).
- Model – The result of applying an algorithm to data (e.g., a trained neural network).
3.4. Evaluation Metrics
| Task |
Common Metric |
| Classification |
Accuracy, Precision, Recall, F1‑score, ROC‑AUC |
| Regression |
Mean Squared Error (MSE), R² |
| Clustering |
Silhouette score, Adjusted Rand Index |
| Generation (NLP/vision) |
BLEU, ROUGE, Inception Score, FID |
4. Popular AI Techniques
| Technique |
Core Idea |
When to Use |
| Linear / Logistic Regression |
Simple linear mapping (or probability) from inputs to output. |
Baseline models, interpretability needed. |
| Decision Trees & Random Forests |
Hierarchical splitting of data; ensembles reduce variance. |
Tabular data, mixed feature types. |
| Support Vector Machines (SVM) |
Finds optimal separating hyperplane with margin maximization. |
High‑dimensional data, limited training samples. |
| Neural Networks (Deep Learning) |
Stacked layers of nonlinear units learn hierarchical representations. |
Images, speech, large text corpora. |
| Convolutional Neural Networks (CNNs) |
Convolutions capture spatial patterns. |
Computer vision, video analysis. |
| **Recurrent Neural Networks (RNNs) & Transformers |
|
|