section

Part 1 — What AI is

Learn the smallest set of AI concepts needed to choose a model, prepare context, and select an adaptation method.

Use these pages when a project decision depends on how a model works. You do not need to complete the part before starting a project.

Lesson	Use it when
What a model actually does	An answer sounds confident and you need to decide what to verify.
How models are made	Training, weights, fine-tuning, or alignment are being confused.
Pick a model for the job	You need a deliberate quality, speed, and cost trade-off.
What the model is holding	A thread is becoming long, expensive, or confused.
Choose the right adaptation	You are deciding between instructions, retrieval, and fine-tuning.

Separate a model's plausible continuation from evidence you can trust.

Distinguish training, weights, fine-tuning, and alignment without treating a model as a searchable document store.

Choose model capability and reasoning effort from task difficulty, evidence, latency, and cost instead of habit.

Build a small working context, understand token-based usage, and prevent long threads from obscuring the current task.

Decide whether a failure needs clearer instructions, retrieved source material, or learned recurring behaviour.