A practical AI learning course
Turn AI Jargon Into Working Knowledge
Learn 15 concepts with a curious Orc, then prove each one on a fresh example. No account or live AI required.
Your progress stays in this browser. If local storage is unavailable, you can still learn in this tab.
Progress is saved only for this browser tab. Keep it open to continue this session.
Your Progress
0 of 15 Understood
Nothing queued for Revisit yet.
Open, With a Suggested Path
Course Map
Start anywhere. The numbered route builds a useful mental model from language to reliable agent systems.
Chapter 1
Language
Not Started Learn the units, instructions, and limits of a model conversation.
- TokensA unit of text a language model reads or produces, often smaller than a word.Not started
- Prompts & instructionsThe instructions and context supplied to a model for a particular generation.Not started
- Context windowsThe bounded amount of token context a model can consider in one request.Not started
Language learning pathA teaching diagram for the 3 concepts in Language. Chapter 2
Model output
Not Started Understand generation, reliable shapes, and unsupported answers.
- Models, LLMs & inferenceA trained model uses learned patterns at inference time to generate token sequences.Not started
- Structured outputModel output constrained to a declared machine-readable schema.Not started
- Hallucination & groundingA fluent model output that is unsupported, fabricated, or otherwise incorrect.Not started
Model output learning pathA teaching diagram for the 3 concepts in Model output. Chapter 3
Knowledge
Not Started Represent meaning, search it, and ground generation.
- EmbeddingsA numerical vector representing aspects of an input so related items can be compared.Not started
- Semantic searchRetrieval that ranks items by learned meaning similarity rather than exact terms alone.Not started
- Retrieval-augmented generationRetrieving relevant external material and supplying it as context for generation.Not started
Knowledge learning pathA teaching diagram for the 3 concepts in Knowledge. Chapter 4
Action
Not Started Link models to capabilities through bounded loops.
- Tools & function callingA structured model request for application code to invoke a defined external capability.Not started
- Model Context Protocol (MCP)An open protocol for connecting AI applications to tools and contextual data through standard interfaces.Not started
- Agents & agent loopsA system that uses a model in a loop to choose actions, observe results, and pursue a goal.Not started
Action learning pathA teaching diagram for the 3 concepts in Action. Chapter 5
Reliability
Not Started Manage state, approval, and measurable quality.
- Memory & stateApplication-managed information carried across steps or sessions, distinct from a model's temporary context.Not started
- Guardrails & human approvalLayered controls that constrain, check, or require confirmation for model-driven actions.Not started
- EvalsRepeatable tests that measure an AI system against defined tasks and criteria.Not started
Reliability learning pathA teaching diagram for the 3 concepts in Reliability.
Newest First
Recent History
- No attempts yet. Your newest 5 lesson and Revisit results will appear here.
Local Learning Artifact
Mastery Certificate
Understand all 15 concepts to unlock a printable certificate. It is not a verified credential.
15 of 15 Understood
Orc Jargon Course Certificate
This local certificate recognises
an AI-curious builder
for demonstrating transfer-level understanding across all 15 course concepts.
Course Data
Clear course progress, Revisit items, and History from this browser. Your theme setting remains unchanged.