AI pillar · Module 1 of 6

What is AI, really?

Forget the Hollywood robots. AI today is basically very sophisticated pattern matching. That’s it. But “pattern matching” turns out to be surprisingly powerful when you do it at massive scale.

← Back to AI Fundamentals Training

1.1 The core idea

Here’s the simplest way I can explain it:

Traditional software: You write rules. “If the customer’s order is over $100, give them free shipping.” The computer follows your rules exactly.

AI: You show examples. “Here are 10,000 emails that are spam, and 10,000 that aren’t. Figure out the pattern.” The computer learns to recognise spam on its own.

This shift—from writing rules to showing examples—is what makes AI different. And it’s why AI can do things that would be impossible to program manually, like recognising faces or understanding natural language.

The flavours of AI you’ll hear about

  • Machine learning (ML). The umbrella term for systems that learn from data. Most AI today is ML.
  • Deep learning. ML using neural networks with lots of layers. This is what powers image recognition, speech recognition, and language models. It’s called “deep” because of the many layers, not because it thinks deeply.
  • Generative AI. The stuff making headlines—ChatGPT, Claude, DALL-E, Midjourney. These systems generate new content (text, images, code) based on patterns they’ve learned.
  • Large Language Models (LLMs). The engines behind chatbots like ChatGPT. They’re trained on massive amounts of text and learn to predict what words come next. That simple idea scales into surprisingly capable systems.

1.2 What AI can and can’t do

Let’s be honest about capabilities:

AI is genuinely good at:

  • Pattern recognition (faces, objects, anomalies)
  • Processing huge amounts of data quickly
  • Generating text, images, and code
  • Translation between languages
  • Specific games (chess, Go, video games)
  • Predictions based on historical patterns

AI still struggles with:

  • True understanding (it mimics understanding)
  • Common sense reasoning
  • Novel situations outside training data
  • Explaining its reasoning reliably
  • Knowing what it doesn’t know
  • Physical world interaction (robotics is hard)

💡 The key insight

AI is a tool, not a mind. It’s incredibly useful for specific tasks, but it doesn’t “understand” anything in the way humans do. It finds patterns and generates outputs based on those patterns. That’s both its power and its limitation.

Free resources to go deeper