What Is an LLM?
An LLM predicts and generates language one token at a time using the context available in the conversation.
A small unit of text.
The information currently available.
Choosing likely next tokens.
An LLM is not a searchable encyclopedia
A language model generates sequences based on learned statistical structure. It can reproduce useful knowledge patterns, but it does not “open a fact record” before every sentence.
This explains why context matters so much: the instructions and evidence you provide can strongly shape the next tokens the model generates.
Quick check
Explain this lesson back in one sentence before continuing. If you cannot, reread the example and key takeaway.
Do not confuse confident language with guaranteed understanding.
LLMs generate language from tokens and context. Better context improves control, but verification still matters.