Humanizing Artificial Intelligence

Role Play Prompting

Role Play Prompting involves instructing the language model to adopt a specific persona or role while generating its response.

By asking the model to “act as” a particular character, profession, or personality type, you can elicit responses with a specific tone, style, and perspective. This technique helps tailor the output to your specific needs and can make interactions more engaging.

Pros:

  • Elicits responses with a specific tone, style, and perspective
  • Makes the output more tailored and engaging for specific applications
  • Can be useful for simulating different viewpoints or expertise
  • Adds personality and character to the responses

Cons:

  • The model might sometimes overemphasize the persona
  • Can produce responses that are less factual or comprehensive if the role conflicts with accuracy
  • May require additional guidance to balance role-playing with factual correctness
  • Not all roles will be equally well-represented in the model’s training data

Example:

Prompt:

Act as a friendly and knowledgeable history tutor explaining the causes of the French Revolution.

Expected Output: (A response explaining the causes of the French Revolution in a friendly and knowledgeable tutor’s voice)

Role play prompting works best for:

  • Educational content with a specific tone
  • Customer service simulations
  • Creative writing with distinct character voices
  • When you want expertise in a specific domain
  • Making technical explanations more approachable

For more subtle guidance on tone and style, directional stimulus prompting offers a less explicit alternative.