🤖 What is Artificial Intelligence?
Artificial Intelligence (AI) is the field of computer science focused on creating systems that can perform tasks that typically require human intelligence—like understanding language, recognizing images, making decisions, and learning from experience.
The AI Hierarchy
Any technique enabling computers to mimic human behavior
Algorithms that improve through experience
Neural networks with many layers
Key Characteristics of AI
- Learning: Improving performance based on experience
- Reasoning: Drawing conclusions from information
- Problem Solving: Finding solutions to complex tasks
- Perception: Understanding sensory input (images, sounds)
- Language: Understanding and generating human language
🧠 Machine Learning
Machine Learning (ML) is a subset of AI where computers learn patterns from data instead of being explicitly programmed. The computer improves its performance on a task by learning from examples.
Machine Learning: Data + Answers → Rules
Types of Machine Learning
📚 Supervised Learning
Learning from labeled examples. The algorithm is given input-output pairs and learns to map inputs to correct outputs.
Examples: Email spam detection, image classification, price prediction
🔍 Unsupervised Learning
Finding patterns in unlabeled data. The algorithm discovers hidden structures without being told what to look for.
Examples: Customer segmentation, anomaly detection, data clustering
🎮 Reinforcement Learning
Learning through trial and error. The algorithm learns by taking actions and receiving rewards or penalties.
Examples: Game playing (Chess, Go), robotics, self-driving cars
- Data: Thousands of emails labeled as "spam" or "not spam"
- Training: Algorithm learns patterns (e.g., words like "FREE", "WINNER")
- Prediction: New emails are classified based on learned patterns
- Improvement: Model gets better as it sees more examples
🕸️ Neural Networks
Neural networks are computing systems inspired by the human brain. They consist of layers of interconnected "neurons" that process information and learn to recognize patterns.
How Neural Networks Work
- Input Layer: Receives the data (e.g., pixels of an image)
- Hidden Layers: Process and transform the data through weighted connections
- Output Layer: Produces the final prediction or classification
Training: The network adjusts its "weights" (connection strengths) based on errors, gradually improving accuracy.
Deep Learning
Deep learning uses neural networks with many hidden layers (hence "deep"). This allows the network to learn increasingly complex features at each layer.
When a deep learning model analyzes a photo of a cat:
- Layer 1: Detects edges and simple shapes
- Layer 2: Recognizes basic parts (curves, corners)
- Layer 3: Identifies features (eyes, ears, whiskers)
- Layer 4: Combines features into "cat" concept
- Output: "This is a cat" (95% confidence)
📊 Types of AI
By Capability
| Type | Description | Example |
|---|---|---|
| Narrow AI (ANI) | Specialized in one task; all current AI | Siri, spam filters, recommendation systems |
| General AI (AGI) | Human-level intelligence across all tasks | Does not exist yet (theoretical) |
| Super AI (ASI) | Surpasses human intelligence in every way | Science fiction (not possible yet) |
By Functionality
| Type | Description | Example |
|---|---|---|
| Reactive Machines | Responds to current input only; no memory | IBM's Deep Blue (chess) |
| Limited Memory | Uses past data to make decisions | Self-driving cars, ChatGPT |
| Theory of Mind | Understands human emotions and intentions | In research (not fully achieved) |
| Self-Aware | Has consciousness and self-awareness | Theoretical (does not exist) |
All AI today is "Narrow AI"—excellent at specific tasks but unable to generalize like humans. ChatGPT can write essays but can't drive a car. Tesla's AI can drive but can't hold a conversation.
🌍 AI Applications
🗣️ Natural Language Processing
- Chatbots (ChatGPT, Claude)
- Translation (Google Translate)
- Voice assistants (Siri, Alexa)
- Sentiment analysis
👁️ Computer Vision
- Facial recognition
- Medical image analysis
- Self-driving vehicles
- Product quality inspection
🎯 Recommendation Systems
- Netflix/YouTube suggestions
- Amazon product recommendations
- Spotify music discovery
- Social media feeds
🎨 Generative AI
- Image generation (DALL-E, Midjourney)
- Text generation (GPT models)
- Code generation (GitHub Copilot)
- Music composition
📚 Key AI Terms
| Term | Definition |
|---|---|
| Algorithm | A set of rules or instructions for solving a problem |
| Training Data | The data used to teach a machine learning model |
| Model | The output of training; the "learned" system that makes predictions |
| Feature | An individual measurable property of the data (e.g., age, color) |
| Accuracy | How often the model makes correct predictions |
| Bias | Systematic errors from flawed assumptions in training data |
| Overfitting | Model memorizes training data but fails on new data |
| LLM | Large Language Model—AI trained on massive text data (e.g., GPT) |
| Prompt | The input/question you give to an AI system |
✏️ Practice Questions
AI is the broadest term—any technique that enables computers to mimic human intelligence.
Machine Learning is a subset of AI where systems learn from data instead of explicit programming.
Deep Learning is a subset of ML that uses neural networks with many layers to learn complex patterns.
Think of it as: AI > ML > Deep Learning (each is contained within the previous).
Supervised Learning would be most appropriate.
Why: The service has labeled data—users' past ratings/watches (input) and whether they liked movies (output). The algorithm can learn patterns from this data to predict preferences for new movies.
Specifically, this is a recommendation system that uses collaborative filtering or content-based filtering.
ChatGPT is Narrow AI because:
- It's specialized for text generation and conversation
- It can't perform tasks outside its training (can't drive a car, see images, or interact with the physical world)
- It lacks true understanding—it predicts likely word sequences, not comprehends meaning
- It has no memory between conversations (starts fresh each time)
- It can't learn new things after training without retraining
General AI would be able to learn and perform ANY intellectual task a human can do.
AI bias occurs when a machine learning model produces systematically unfair results due to flawed assumptions or unrepresentative training data.
Example: A hiring algorithm trained on historical data from a company that predominantly hired men might learn to favor male candidates—not because men are better, but because the training data was biased.
How it happens:
- Data bias: Training data doesn't represent all groups equally
- Historical bias: Data reflects past discrimination
- Selection bias: Data collected in a non-random way
Solution: Diverse, representative training data, regular bias audits, and human oversight.
⚖️ AI Ethics & Considerations
- Privacy: AI systems often require large amounts of personal data
- Bias: AI can perpetuate or amplify existing biases in data
- Job Displacement: Automation may change or eliminate certain jobs
- Accountability: Who is responsible when AI makes mistakes?
- Misinformation: Generative AI can create convincing fake content
- Verify AI-generated information with reliable sources
- Don't share sensitive personal data with AI systems
- Understand AI's limitations—it can make mistakes
- Use AI as a tool to enhance human work, not replace critical thinking
Explore AI in Your Learning
Centauri uses AI to help you plan, organize, and achieve your goals.
Get Early Access