SHORTEN Prompt

Prompt Engineering - Generating Concise Summaries and Responses

In many situations, we need information to be brief and to the point. Large Language Models (LLMs) can be prompted to generate shorter, more concise responses or summaries of longer texts. This technique improves efficiency in communication and helps users quickly grasp key information.

Understanding Prompts for Conciseness

Requesting concise output from an LLM involves clearly instructing it to shorten information, summarize text, or adhere to specific length constraints. There isn’t a special ‘SHORTEN directive’; effectiveness comes from precise instructions.

A basic interaction might be:

User: Can you explain the concept of artificial intelligence in just one sentence?
Assistant: Artificial intelligence (AI) is a field of computer science focused on creating machines that can perform tasks typically requiring human intelligence, such as learning, problem-solving, and decision-making.

In this example, the user explicitly asks for a one-sentence explanation. The LLM condenses the core idea of AI into a single, coherent sentence.

Best Practices for Requesting Concise Output

To effectively prompt an LLM for conciseness, consider the following best practices:

• Be Explicit About the Desired Output: Clearly state that you want a summary, a brief explanation, key takeaways, or a shortened version. Use phrases like ‘Summarize the following text…’, ‘Explain briefly…’, ‘In a few words…’, ‘Provide a one-sentence summary of…’.
• Specify the Input (if applicable): If you’re summarizing existing text, provide that text clearly within the prompt. For example: ‘Summarize this article for me: [paste article text here].’
• Define Length Constraints: If you have a specific length in mind, state it. Examples: ‘…in no more than 50 words,’ ‘…in three bullet points,’ ‘…as a tweet (under 280 characters).’
• Focus on Key Information vs. Truncation: Emphasize that the goal is to extract the most important information, not just to cut off the text. You can ask: ‘What are the main points of…?’ or ‘Extract the key findings from this report.’
• Maintain Core Meaning: Stress that conciseness should not sacrifice the essential meaning or accuracy of the information unless you explicitly ask for a very high-level overview where some nuance might be lost.
• Iterate if Necessary: If the first attempt isn’t concise enough or misses key details, refine your prompt. You might say, ‘That’s good, but can you make it even shorter?’ or ‘Please focus more on [specific aspect] in your summary.‘

Example Application: Python Implementation

Let’s explore a Python example using the OpenAI API to request a concise explanation of a complex topic, like Einstein’s theory of relativity.

from openai import OpenAI

client = OpenAI(api_key='YOUR_API_KEY') # Ensure API key is set

def get_concise_explanation(topic, length_constraint):
    try:
        prompt = f"Explain the core idea of '{topic}' {length_constraint}. Focus on the most essential aspects."
        response = client.chat.completions.create(
            model="gpt-3.5-turbo",
            messages=[
                {"role": "system", "content": "You are an AI assistant skilled at providing clear and concise explanations."},
                {"role": "user", "content": prompt}
            ],
            max_tokens=100,  # Adjust based on expected length of concise output
            temperature=0.3, # Lower temperature for more factual and direct responses
            n=1,
            stop=None
        )
        return response.choices[0].message.content.strip()
    except Exception as e:
        return f"An error occurred: {e}"

topic_to_explain = "Einstein's theory of general relativity"
constraint = "in one or two simple sentences"

explanation = get_concise_explanation(topic_to_explain, constraint)
print(f"Concise explanation of {topic_to_explain}:")
print(explanation)

In this example, get_concise_explanation() constructs a prompt specifying the topic and a clear length constraint.

Example Output

Running the script might produce an output like this:

Concise explanation of Einstein's theory of general relativity:
Einstein's theory of general relativity describes gravity not as a force, but as a curvature of spacetime caused by mass and energy, which dictates how objects move through it.

(Another run might produce a slightly different, but still concise, valid explanation.)

Conclusion

In this chapter, we explored how to prompt LLMs to generate concise summaries and responses. By clearly specifying the need for brevity, defining length constraints, and focusing on the extraction of key information, you can effectively use LLMs to distill complex topics into easily digestible snippets. This skill is invaluable for efficient communication and quick understanding in a fast-paced world.