LLM Beginner Guide: prompts, tokens, context and hallucinations

A language model does not look up one fixed answer. It predicts a useful continuation from your input and its training patterns.

A strong prompt usually combines role, context, task, constraints, examples and output format.

What you will learn

Level 1What is an LLM?Level 1What is a prompt?Level 1What are tokens?Level 1Why does AI sometimes invent things?Level 1Why does context help?Level 1Search engine vs. LLM Level 1What does “AI understands language” mean?

What is an LLM?

An LLM, or large language model, is software trained on huge amounts of text to predict and generate useful language. The practical way to think about it: you give it a task, it turns your text into tokens, estimates likely next tokens and returns a draft. It can write, summarize and classify because many language patterns are compressed into its parameters. It is not a database and it does not automatically know whether a fresh fact is true.

Example: A SaaS support lead asks: “Summarize this complaint and propose a calm refund reply.” The model does not “feel” the complaint; it recognizes the pattern of complaint, refund policy and professional tone.

What is a prompt?

A prompt is the instruction package you send to an AI system. Good prompts usually include the role, the situation, the exact task, constraints, examples and the desired output format. A weak prompt asks for “ideas”. A strong prompt says who the ideas are for, what they must achieve and how they should be judged.

Example: Weak: “Write a product description.” Better: “Act as a DTC ecommerce copywriter. Write a 90-word product description for a reusable coffee cup. Audience: commuters. Tone: practical, not hype. Include one headline and three bullets.”

What are tokens?

Tokens are the small text units a model reads and writes. A token can be a whole word, part of a word, punctuation or a character, depending on the tokenizer and language. Token limits matter because input plus output must fit inside the model context window.

Example: “PromptingEasy helps teams” might become tokens similar to “Prompt”, “ing”, “Easy”, “helps”, “teams”. This is why long documents and many examples increase cost and may crowd out important instructions.

Why does AI sometimes invent things?

AI can hallucinate when it generates a plausible-sounding continuation without enough reliable grounding. The model is optimized to produce likely text, not to guarantee truth by default. Hallucinations become more likely when the question asks for obscure facts, fresh information, hidden data or citations that were not provided.

Example: A user asks for “the exact 2026 price of a niche API plan” without browsing or source text. The model may produce a confident-looking price because that shape of answer is common, even if the number is wrong.

Why does context help?

Context narrows the search space. If the model knows the audience, goal, constraints, examples and source material, it can produce an answer that fits your situation instead of a generic answer. More context is not always better: irrelevant context can distract the model and increase cost.

Example: A marketing team gets better ad copy when it includes the product positioning, target customer, banned claims and two successful past ads instead of only saying “write ads”.

Search engine vs. LLM

A search engine retrieves pages and ranks links. An LLM generates an answer from patterns and provided context. Search is better when you need current sources, official pages or multiple perspectives. LLMs are better when you need synthesis, rewriting, reasoning over supplied information or structured drafts. Many strong AI products combine both.

Example: For “latest tax deadline in Zurich”, use search or official sources. For “turn these notes into a polite customer email”, an LLM is the better interface.

What does “AI understands language” mean?

In everyday language, “understands” means the model can respond appropriately to meaning, tone and structure. Technically, it learned statistical representations that map text patterns to useful outputs. It does not understand like a person with lived experience, intentions or common-sense accountability.

Example: If you write “make this less salesy,” the model can often adjust tone because it has learned patterns of salesy vs. neutral language. That is useful linguistic competence, not human understanding.

Mini test you can run

Pick five real tasks from your own workflow. Run one short prompt, one structured prompt and one prompt with examples or source context. Score each output from 1 to 5 for usefulness, factual risk and edit time. Keep the winning prompt as your baseline and retest after every major change.

Variant	Usefulness	Factual risk	Edit time
Short prompt	Medium	Higher	High
Structured prompt	High	Medium	Medium
Context + examples	Highest for repeat tasks	Lower if sources are good	Low

Back to AI Wiki overview Read the prompting tutorials