1.1.2. AI vs. Machine Learning vs. Deep Learning
💡 First Principle: These three terms describe ever-narrower circles. AI is "make software act intelligently." Machine learning narrows that to "make software learn the rules from examples instead of being told them." Deep learning narrows further to "use many-layered neural networks to learn those rules from very large, complex data." Each step trades broad applicability for greater power on hard perception and language problems.
The key shift from traditional programming to machine learning is where the rules come from. In traditional code, a developer writes the rules. In machine learning, you feed the system labeled examples and it derives the rules itself — which is why ML shines at problems too messy to write rules for, like recognizing cats in photos. Deep learning is the subset of ML that uses neural networks with many layers; it powers most of today's vision, speech, and generative AI, but it needs far more data and compute than simpler ML.
| Term | What It Means | Rules Come From | Typical Use |
|---|---|---|---|
| Artificial Intelligence | Software doing tasks needing human-like intelligence | Developer OR learned | The broad goal; includes rule-based systems |
| Machine Learning | Systems that learn patterns from data | Learned from examples | Predictions, classification, recommendations |
| Deep Learning | ML using many-layered neural networks | Learned from large datasets | Vision, speech, language, generative AI |
⚠️ Exam Trap: Deep learning is a subset of machine learning, which is a subset of AI — not three parallel categories. If a question implies deep learning and machine learning are separate, unrelated fields, that's the wrong framing.
Reflection Question: Why does deep learning need so much more data than a simple machine-learning model? Think about what "many layers" are learning that a simpler model isn't.
