Data, Patterns & Models
📋 What it is
A “model” is the pattern an AI has learned from data; better, more varied data makes a better model.
🗣️ Coach says
After learning, the pattern the AI keeps is called a MODEL. The rule is simple: garbage in, garbage out. If the training data is limited or biased, the model is too. Good, varied data → a model that works fairly for everyone.
🧠 Memory hook
A model = the learned pattern. Garbage data in → garbage model out.
😂 Giggle
Why did the password go to therapy?
It had too many issues with trust and validation!
😲 Whoa!
An early AI meant to spot skin problems worked poorly on darker skin — because its training photos were mostly of light skin. The data shaped the model.
✅ Quick check: An AI trained only on photos of sunny days is shown a snowy scene. What might happen?
Say your answer out loud first — then reveal.
It may fail — it never learned snow; a model only knows the patterns in its training data.
A model can only handle what its data taught it; gaps in data become gaps in the model.
🧪 Try it! (2 minutes)
Imagine training a “fruit AI” on only apples. Would it recognise a banana? (No — data limits the model.)