AI Vocabulary: 25 Terms Every Beginner Should Know
AI is full of jargon, and the jargon is a barrier. Once you know the vocabulary, every article, product page and news story becomes readable. This glossary demystifies the 25 most important AI terms — one clear line each.
Skim it, bookmark it, and come back whenever a term trips you up. We’ve grouped the terms so related concepts sit together.
- A handful of terms explain most AI behavior: token, context window, parameters, inference.
- Data terms: embedding, vector database, RAG, fine-tuning.
- Capability terms: multimodal, AGI, ASI, hallucination.
- Knowing the vocabulary makes every other AI topic easier.
- You can test all of it in the AI Vocabulary quiz category.
Want to check your understanding as you read? You can take our free AI Quiz any time — it covers this topic across Beginner, Intermediate and Advanced levels.
Core model terms
The words that explain how a model processes text:
- Token — a chunk of text (word or word-piece) a model processes.
- Context window — how much text (in tokens) a model can consider at once.
- Parameters — the learned internal weights that store what a model knows.
- Inference — running a trained model to get outputs (as opposed to training it).
- Prompt — the input or instruction you give a model.
- Temperature — a setting controlling randomness/creativity of output.
Data & retrieval terms
How AI understands meaning and uses your data:
- Embedding — text turned into a numeric vector that captures meaning.
- Vector database — stores embeddings for fast similarity search.
- RAG — Retrieval-Augmented Generation; fetching data to ground answers.
- Fine-tuning — further training a model on specific data.
- Training data — the examples a model learns from.
- Quantization — reducing numeric precision to make models smaller/faster.
Capability & type terms
What models can do, and the big-picture categories:
- Multimodal — handles multiple data types (text, image, audio).
- Generative AI — creates new content rather than only classifying.
- AGI — Artificial General Intelligence; human-level breadth (not yet real).
- ASI — Artificial Superintelligence; beyond humans (hypothetical).
- Hallucination — a confident but false model output.
- Alignment — making AI act in line with human intentions.
🧠 Test what you just learned
Put your knowledge to the test with our free 250-question AI Quiz — 14 categories, instant explanations, a grade and a shareable certificate.
Building & deploying terms
The words you’ll meet when AI meets real software:
- API — an interface that lets software talk to an AI service.
- Latency — the delay before a response.
- Throughput — how much work (tokens/requests) is handled over time.
- Benchmark — a standardized test comparing model performance.
- Open-weight — a model whose weights are publicly downloadable.
- RLHF — Reinforcement Learning from Human Feedback, used to align models.
- MCP — Model Context Protocol, a standard connecting models to tools.
Related reading
Frequently asked questions
What’s the most important term to learn first?
‘Token’ and ‘context window’ — they explain how models read text and why length limits and per-token pricing exist.
What’s the difference between AGI and today’s AI?
Today’s AI is narrow (task-specific). AGI would match humans across most tasks — and it doesn’t exist yet.
Is ‘generative AI’ the same as ‘AI’?
No — it’s a subset that creates new content. AI is the broader field.
Do I need to memorize all these?
No. Understanding the core few (token, context window, hallucination, embedding, RAG) covers most conversations.
Where can I test these terms?
The AI Vocabulary category in our free AI Quiz is built exactly for this.
🧠 Test what you just learned
Put your knowledge to the test with our free 250-question AI Quiz — 14 categories, instant explanations, a grade and a shareable certificate.
Keep going: take the free AI Quiz and see if you rank as an AI Master.
