Role Prompting: Definition and Examples

Role prompting is a prompt engineering technique that consists of asking a language model to adopt a specific role before responding to a query. By assigning an identity — for example, "You are a cybersecurity expert" or "Act as a physics teacher" — the model is conditioned to draw on the knowledge and language register associated with that character.

This technique is based on a simple principle: language models have been trained on massive corpora containing texts produced by professionals in all fields. By specifying a role, we activate a subset of this knowledge, allowing for more relevant, precise, and well-calibrated responses for the intended context.

Role prompting can be used simply (a single instruction sentence) or sophisticatedly, by combining the role with an experience level, target audience, format constraints, and situational context. For example, asking the model to be "a senior Python developer specialized in data engineering, who explains to a junior" produces very different results from "a theoretical computer science researcher."

This approach is particularly effective for creative, educational, technical, or writing tasks. It often forms the first building block of a well-structured prompt and naturally combines with other techniques like few-shot prompting or chain-of-thought.