Generative AI & LLMs
Next-token prediction
The training objective where a language model learns to predict the next token in a sequence.
Example
Given “The meeting starts at,” the model assigns probabilities to possible next tokens.
Why it matters
It gives systems a way to learn language patterns from large amounts of writing.
You'll hear it
“That phrase is a likely continuation, but it isn't the right fact.”
Can you control it?
Not directly
No, not directly. Next-token prediction describes a learned property, observed result, or broader concept rather than a normal user setting. You may influence it indirectly by changing the tool, data, prompt, or workflow.
Common questions
- Does the most likely next word have to be true?
- No. Familiar wording can be a poor match for the real situation.
- Does it consider only the immediately preceding word?
- No. The available earlier text can affect the choice, within the system's limits.
- Can this simple learning task support complex behavior?
- Yes. Learning many language patterns can support broader abilities, although their reliability still varies.