Step 1 of 16
A clear mental model of what AI can and cannot do — set realistic expectations before diving into tools.
สร้างความเข้าใจที่ถูกต้องว่า AI ทำอะไรได้ ทำอะไรไม่ได้ และเหมาะกับอะไร — ก่อนจะเริ่มใช้งาน
Artificial Intelligence (AI)
├── Machine Learning (ML) — learns from data, no explicit rules
│ ├── Supervised Learning — learns from labeled examples
│ ├── Unsupervised Learning — finds patterns on its own
│ └── Deep Learning — neural networks with many layers
│ ├── Computer Vision — recognizes images, faces, objects
│ ├── NLP — understands and generates language
│ └── Generative AI — creates NEW content (text, images, audio)
Each level builds on the one before. Generative AI is a subset of Deep Learning, which is a subset of Machine Learning, which is a subset of AI.
The broad field: making computers do things that normally require human intelligence. This includes rule-based systems, expert systems, robotics — not just the modern stuff.
Instead of programming explicit rules, you feed data and the system learns patterns on its own.
Examples you use daily:
These are predictive AI — they predict an outcome (spam/not spam, fraud/legit, what you'll buy) based on patterns in data. They don't create anything new. They classify, predict, and recommend.
ML using neural networks with many layers. Enables more complex tasks like recognizing faces, understanding speech, and driving cars.
Examples:
Deep learning models that create NEW content — text, images, code, audio, video — by learning patterns from massive datasets and predicting what comes next.
Examples:
| Predictive AI | Generative AI | |
|---|---|---|
| What it does | Classifies, predicts, recommends | Creates new content |
| Output | A label, a number, a ranking | Text, image, audio, code |
| Training | Learns from labeled data | Learns from massive text/image datasets |
| Example | "This email is spam" (classification) | "Write me a professional email" (generation) |
| Example | "You might also like this product" (recommendation) | "Create an image of a cat in a business suit" (generation) |
| Risk | Wrong prediction | Plausible but wrong information (hallucination) |
| You've used it since | ~2010 (spam filters, recommendations) | ~2023 (ChatGPT, Claude, Gemini) |
Key insight: You've been using predictive AI for 10+ years. Generative AI is what's new — and it's why everyone is talking about AI now.
What it DOES:
Every AI interaction involves four skills:
| Skill | What It Means |
|---|---|
| Delegation | Deciding what to ask AI to do (and what NOT to) |
| Description | Writing clear instructions so AI understands your intent |
| Discernment | Evaluating AI's output — is it correct, useful, safe? |
| Diligence | Using AI responsibly — privacy, ethics, long-term impact |
You'll practice all four throughout this course.
| Type | Examples | Best For |
|---|---|---|
| Chat AI | Claude, ChatGPT, Gemini | Writing, analysis, brainstorming, coding |
| AI Search | Perplexity, Google AI Overview | Finding current information with sources |
| AI Agents | Claude Cowork, Claude Code, OpenAI Operator | Multi-step tasks that take minutes, not seconds |
| Image AI | Gemini Imagen, DALL-E, Midjourney | Creating images from text descriptions |
| Voice AI | Google NotebookLM, ElevenLabs | Audio summaries, voice generation |
| Predictive AI | Gmail spam filter, Shopee recommendations | Classification, prediction, personalization |
AI is not one thing — it's a family of technologies. You've used predictive AI (spam filters, recommendations) for years. Generative AI (text, image, audio creation) is the new layer. Understanding the difference helps you set realistic expectations: predictive AI classifies, generative AI creates, and both need human oversight.