Skip to main content
Practice

Zero Shot Prompting: Asking without Examples

Chapter 2 introduces key terms and techniques that often appear in prompt engineering.

If you are interested in AI, you might have heard terms like Zero-Shot/Few-Shot Prompting, and the next-generation search method called RAG (Retrieval Augmented Generation). Here, we cover various terms and techniques related to prompt engineering.

First, let's explore the most basic technique in prompt engineering, Zero Shot Prompting.


What is Zero Shot Prompting?​

In artificial intelligence learning, a shot refers to an example. Therefore, Zero-Shot means there are no examples, i.e., the AI processes a new task based on previously learned data without additional specific examples.


Examples of Zero Shot Prompts​

Example of Zero Shot Prompt 1
Explain the difference between aerobic and anaerobic exercise in less than 200 words.
Example of Zero Shot Prompt 2
Tell me the highest mountain in the world and its height.

As shown above, making the AI interpret and produce results based solely on pre-trained data is called Zero Shot Prompting.

Most everyday prompts without specific examples are classified as Zero Shot Prompts.

Note: Prompts that include a few examples are called Few-Shot Prompts.


Example of Few Shot Prompt
Here are a few examples of news articles. Each article is categorized into a specific category.

Article: "Trade negotiations between countries are making progress. Economic experts view this positively."
Category: Economy

Article: "A famous actor has announced their appearance in a new movie. Fans are thrilled."
Category: Entertainment

Now, categorize the article below.

Article: "Recent research suggests that a healthy diet can reduce the risk of heart disease."
Category:

We will cover Few-Shot Prompting in more detail in the next lesson.


What are the advantages of Zero Shot Prompting?​

  • Quick Results: Minimizes the time spent on drafting the prompt, and allows rapid utilization of the existing model without additional training for new fields or tasks.

  • Flexibility: Not restricted to specific examples or data, it can flexibly adapt to the user's prompt and perform tasks in various fields.


What are the disadvantages?​

  • Accuracy and Expertise Issues: The answer accuracy might drop for tasks in fields that have not been learned yet.

  • Time Wastage Due to Ambiguity: Without clear guidelines, it can be hard for the AI to provide the correct answer. This can require more interaction between the user and the AI, resulting in unnecessary time consumption.


Practice​

Send prompt examples and compare the AI's responses.

Want to learn more?

Join CodeFriends Plus membership or enroll in a course to start your journey.