Based on the case study below, design a detailed prompt that a business owner could provide to a general-purpose Large Language Model to ensure it responds to customer inquiries appropriately.

Google

An appealing feature of prompting is that it allows users to create customized systems. By designing their own unique prompts, individuals can tailor the behavior of a general-purpose Large Language Model to their specific needs and applications without needing to modify the underlying model.

User Customization of LLMs via Prompt Design

A user wants a general-purpose language model to act as a 'Recipe Suggester' that only provides vegan recipes. When the user inputs 'What should I make for dinner?', the model suggests a chicken dish. Which of the following revised inputs would be most effective at customizing the model's behavior to consistently meet the user's specific need?

Customizing an LLM for Customer Service

A company wants to use a general-purpose language model to generate enthusiastic marketing copy for a new eco-friendly water bottle targeted at young adults. The copy must highlight three features: it is BPA-free, insulated, and made from recycled materials.

Consider the two prompts below designed for this task:

**Prompt A:** 'Write marketing copy for a new water bottle.'

**Prompt B:** 'You are a marketing expert writing for a Gen Z audience. Generate an enthusiastic and catchy ad copy for our new eco-friendly water bottle. Make sure to highlight these key features: it's BPA-free, keeps drinks cold for 24 hours with its double-wall insulation, and is made from 100% recycled materials.'

Evaluate which prompt is more likely to produce the desired output. In your response, analyze the specific elements of each prompt and explain how they contribute to (or fail to contribute to) customizing the model's behavior for this specific application.

Evaluating Prompt Effectiveness for Customization

A popular approach to utilizing Large Language Models is to assign them a specific role or persona to adopt during response generation. This technique guides the model to produce answers that align with the characteristics, tone, and perspective expected of the assigned role.

Learn Before

Related