Temperature
A setting that controls the randomness and creativity of AI-generated outputs.
Definition
Temperature is a parameter that adjusts how random or deterministic an AI model's outputs are. Low temperature (near 0) makes outputs more focused and predictable; high temperature (near 1 or higher) makes outputs more varied and creative.
Think of it as a dial between "give me the most likely answer" and "surprise me with something unexpected."
Why It Matters
Matching temperature to your task improves results. Factual questions need low temperature for accuracy; creative brainstorming benefits from higher temperature for variety.
Understanding temperature helps configure AI applications appropriately—you don't want creative fiction from your data analyst or formulaic outputs from your copywriter.
Examples in Practice
A code generation tool uses temperature 0.2 to produce reliable, predictable syntax that's less likely to contain creative but broken solutions.
A brainstorming assistant uses temperature 0.9 to generate diverse, unexpected ideas even if some are impractical.
A documentation system varies temperature by section: low for technical specifications, moderate for explanatory prose.