What is AI?

One-Line Understanding of AI

AI = Artificial Intelligence

Making computers think and act like humans:

Traditional Programs:
Input → Rules written by programmer → Output

AI Programs:
Input → Rules learned by AI itself → Output

AI vs Traditional Programs

Traditional Programs: Fixed Rules

Rules written by programmers:
if (there is a nose in the image) {
  if (there are two eyes) {
    if (there is a mouth) {
      return "face";
    }
  }
}

Problem: Cannot handle complex situations

AI: Learns Rules Itself

AI learns patterns from massive data:

Training stage:
Input: 100 million images (some with cats, some without)
Output: AI learns to identify cats by itself

Application stage:
Input: A new image
Output: AI decides "cat" or "not cat"

Types of AI

1. Narrow AI

Already implemented: Complete specific tasks

Examples:
- Chess AI (AlphaGo)
- Translation AI (Google Translate)
- Image recognition AI (Face ID)
- Voice assistants (Siri, Alexa)

Feature: Can only do one specific thing

2. AGI (Artificial General Intelligence)

Being explored: Think like humans

Goals:
- Can do anything humans can do
- Has general understanding
- Has autonomous consciousness

Status: Not yet achieved

AI Core Capabilities

1. 🎯 Recognition

Computer vision:
- Face recognition (phone unlock)
- Object recognition (self-driving)
- Text recognition (OCR)
- Medical imaging analysis

Speech recognition:
- Speech to text
- Voice to text
- Music recognition

2. 💬 Generation

Text generation:
- Write articles, code
- Translation
- Conversation

Image generation:
- AI draws pictures (Midjourney)
- AI generates videos
- AI edits photos

Audio generation:
- AI voiceover
- Music generation

3. 🧠 Reasoning

Logical reasoning:
- Math problem solving
- Chess/Go
- Route planning

Decision making:
- Recommendation systems
- Self-driving decisions
- Risk assessment

AI Development Timeline

1950s AI concept born
  ↓
1990s Machine learning rises
  ↓
2010s Deep learning breakthrough
  ↓
2020s Large language models explode (GPT, Claude...)
  ↓
2026   AI + Blockchain integration (PulsePay...)

What are Large Language Models (LLM)?

One-Line

LLM = Large Language Model

A massive "text prediction" machine

Training method:
Show it the entire internet's text
Let it learn "what usually follows this text"

Example:
Input: "Today's weather is"
Output: "great" (learned from prediction)

Next Steps

• Large Language Models — Deep understanding of LLM principles
• AI Agent — AI autonomously executes tasks
• AI Pricing — Understand AI cost structure