Google

To elicit a more specialized and expert response from an LLM, a role-playing prompt can be enhanced by providing a detailed persona. This involves specifying not just a profession, like a psychologist, but also their specific field (e.g., developmental psychology in children), research focus (control and intelligence), and a credible background that includes relevant research, publications, and experiments on the topic at hand, such as delayed gratification.

Example of a Detailed Persona in a Role-Playing Prompt

A role-playing prompt for an LLM combines a persona definition with a user query to elicit a specialized response. For example, a user can instruct the model to adopt the persona of a developmental psychologist who is an expert on delayed gratification, citing their specific research background, publications, and experiments. Following this detailed persona setup, the user can then ask a targeted question relevant to that persona, such as, 'Please explain what delayed gratification is?'

Example of a Complete Role-Playing Prompt

A user wants to generate a detailed analysis of how autonomous delivery drones might change city infrastructure. They use the following instruction:

'You are an expert in city planning. Explain the potential impact of autonomous delivery drones on urban infrastructure.'

Critique the effectiveness of the persona ('an expert in city planning') in this instruction. Then, propose specific, detailed improvements to this persona to elicit a more specialized and insightful response. Justify why your proposed changes would lead to a higher-quality answer.

Critique and Refinement of a Prompt Persona

Analyze the following scenario. A user's initial prompt to a language model is provided, along with the model's generic response. Based on the principles of crafting effective personas, identify the weaknesses of the initial prompt and describe three specific, detailed elements you would add to the persona to elicit a more specialized and expert-level explanation.

Improving a Generic LLM Prompt

A user wants a language model to provide a detailed, expert-level explanation of the economic principles behind carbon pricing. Which of the following initial instructions would be most effective at establishing a persona for the model to achieve this goal?

Learn Before

Related