AI Dictionary › Prompting
Prompt priming is the technique of preparing the model with introductory information, examples or instructions before posing the actual request, in order to steer its behavior and the style of the response. It works like a warm-up: it does not yet ask for the final result, but sets the ground on which the model will reason.
In practice, priming means inserting at the start of the prompt a context, a tone, technical vocabulary or a short example that the model should mirror in its subsequent response. As the model processes text sequentially, it tends to maintain stylistic and conceptual consistency with what it read just before, an effect that stems from how attention weighs preceding text during generation.
In concrete use, priming is used to make the model adopt a specific linguistic register before writing a text, to introduce domain terminology before a technical analysis, or to show a short style excerpt the model should imitate in a longer text. It is a cross-cutting technique that often precedes or accompanies other strategies such as role prompting or few-shot.
The term "priming" is borrowed from cognitive psychology, where it denotes exposure to a stimulus that influences a subsequent response without the person being aware of it. Applied to language models, the concept was adopted to describe how the initial text of a prompt conditions the model's behavior throughout the rest of the generation.
Grace evaluates whether you open your prompts with good context or terminology priming when the scenario calls for a specific linguistic register.
From our network
AGORÀ Intelligence: Enterprise AI Governance Platform
Govern AI at scale: policies, adoption and measurable results on your data. Built for boards and C-suite.
Visit agora-intelligence.com →From the Agora Intelligence blog
📱 Download the Android app (beta) iOS coming soon
Say what you mean. Get what you need.
Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.