ACT LIKE Prompt

Prompt Engineering: Instructing LLMs to ‘Act Like’ a Persona

Modern Natural Language Processing (NLP) models, such as those powering ChatGPT, can generate remarkably human-like text. A key technique in prompt engineering to guide their responses is instructing the model to ‘act like’ a specific character, person, or entity. This is often referred to as persona prompting or role-playing.

This chapter explores the concept of ‘Act Like’ prompts, provides examples, and highlights their applications in various scenarios to tailor the LLM’s behavior, tone, and knowledge.

Understanding ‘Act Like’ Prompts

• Definition: An ‘Act Like’ prompt explicitly instructs the LLM to adopt a specific persona, role, or character when generating its response. This goes beyond just answering a question; it’s about embodying an identity.
• Role-Playing: These prompts enable users to engage with the model in a more immersive, targeted, or creative way by having it simulate different personas.
• Influencing Responses: By specifying a character, profession, or even an abstract concept (e.g., ‘Act like a cautious optimist’), users can significantly influence the model’s language style, tone, vocabulary, and the domain of knowledge it draws upon. For instance, asking it to act like a pirate will result in different language than asking it to act like a theoretical physicist.

Examples of ‘Act Like’ Prompts

Here are a few examples demonstrating the versatility of this technique:

Acting as a Historical Figure:

• Prompt: “You are to act as Leonardo da Vinci. Describe your observations about the flight of birds and how it might inspire a flying machine.”
• Expected Response: The model would attempt to generate a response in a style and with knowledge consistent with Leonardo da Vinci, focusing on observation, mechanics, and inventive thinking.

Impersonating a Fictional Character:

• Prompt: “Assume the persona of Sherlock Holmes. A valuable painting has vanished from a locked gallery with no apparent entry points. What are your initial thoughts and questions?”
• Expected Response: The model would adopt Holmes’ characteristic deductive reasoning, attention to detail, and inquisitive tone.

Simulating an Expert or Professional:

• Prompt: “Act as a seasoned financial advisor. Explain the concept of dollar-cost averaging to a novice investor, highlighting its benefits and potential drawbacks.”
• Expected Response: The model would provide a clear, professional explanation, using appropriate financial terminology but making it accessible to a beginner.

Applications of ‘Act Like’ Prompts

• Creative Writing and Storytelling: Writers can use ‘Act Like’ prompts to generate dialogue, character monologues, or narrative passages in the voice of specific characters, adding depth and authenticity.
• Education and Learning: Students can interact with LLMs role-playing as historical figures, scientists, or literary characters to gain a more immersive understanding of different subjects.
• Content Generation: Marketers can use this to generate content in a specific brand voice or for a particular target audience persona.
• Brainstorming and Problem-Solving: Adopting different expert personas can help explore a problem from multiple angles (e.g., “Act like an engineer, then act like a designer, to propose solutions for… ”).
• Entertainment and Games: ‘Act Like’ prompts are fundamental in creating interactive experiences for chat-based games or virtual assistants, where users engage with distinct virtual characters.

Example: Python Implementation with OpenAI API

Let’s see how to implement an ‘Act Like’ prompt using Python and the OpenAI API.

from openai import OpenAI
import os

# client = OpenAI(api_key=os.environ.get("OPENAI_API_KEY"))
client = OpenAI(api_key='YOUR_OPENAI_API_KEY') # Replace with your key

def generate_persona_response(persona, query):
    try:
        response = client.chat.completions.create(
            model="gpt-3.5-turbo",
            messages=[
                {
                    "role": "system",
                    "content": f"You are to act as {persona}. Respond to the user's query in character, maintaining the persona's typical tone, style, and knowledge."
                },
                {
                    "role": "user",
                    "content": query
                }
            ],
            max_tokens=250, # Adjust based on expected response length
            n=1,
            stop=None,
            temperature=0.7 # Allows for some creativity while staying in character
        )
        return response.choices[0].message.content.strip()
    except Exception as e:
        return f"An error occurred: {e}"

# Example usage:
persona_to_adopt = "Sherlock Holmes"
mystery_query = ("A valuable diamond necklace has been stolen from a locked room. "
                 "The room has no windows, and the only entrance is a solid wooden door, "
                 "locked from the inside. There are no signs of forced entry. "
                 "The owner claims nobody else had access. How might the thief have managed this?")

# Using a more direct prompt within the query can also reinforce the persona
full_query_for_sherlock = f"Mr. Holmes, consider this perplexing case: {mystery_query} What are your deductions?"

sherlock_response = generate_persona_response(persona_to_adopt, full_query_for_sherlock)
print(f"Response from {persona_to_adopt}:\n")
print(sherlock_response)

In this Python code, the generate_persona_response function takes a persona and a query. The system message sets up the persona for the LLM. The user’s query is then processed by the LLM acting in that role.

When running this, the LLM’s response should reflect Sherlock Holmes’s analytical and observant nature:

Response from Sherlock Holmes:

"Intriguing, my dear Watson, positively intriguing! A classic locked-room mystery, if ever I saw one. Let us dissect the elements before us.

First, 'locked from the inside' and 'no signs of forced entry' are key. This immediately suggests several possibilities, ruling out brutish methods.

1.  **The Owner is Not Being Entirely Truthful:** This is always a primary consideration. Could the owner have staged the theft for insurance purposes or some other motive? We must not dismiss this.
2.  **A Hidden Mechanism or Passage:** While the room is described as having 'no windows' and a 'solid wooden door,' one must investigate thoroughly. Are there old chimneys, dumbwaiters, or cleverly concealed panels? The term 'solid' can be deceptive.
3.  **The Thief Never Left, or Was Never 'In' in the Conventional Sense:** Could the thief have hidden within the room prior to it being locked? Or, perhaps, the necklace was removed via some ingenious contraption from outside, through a small, overlooked aperture – a keyhole, a gap under the door, a loose floorboard.
4.  **The 'Locked from the Inside' is a Deception:** Was the door truly locked from the inside *after* the theft? Or was it locked in a way that *appears* to be from the inside? There are mechanisms – strings, specific tools – that could achieve this.
5.  **Collusion:** If the owner claims 'nobody else had access,' this must be verified. Could a servant, or someone with a duplicate key, be involved, and the 'locked from inside' part of a more elaborate ruse?

My immediate line of inquiry would be:
*   A meticulous examination of the room itself – every inch of the walls, floor, ceiling, and the door mechanism.
*   The nature of the lock: Is it simple or complex? Could it be manipulated?
*   The owner's character, financial situation, and recent behavior.
*   The exact timing of when the necklace was last seen and when it was discovered missing.

The game is afoot! We must eliminate the impossible, and whatever remains, however improbable, must be the truth."

Note: The exact response will vary with each generation and model version, but the style and reasoning should align with the requested persona.

Conclusion

‘Act Like’ prompts are a potent tool in the prompt engineer’s toolkit. They allow for fine-grained control over an LLM’s output, enabling users to generate text that is not only informative but also stylistically appropriate for a wide range of creative, educational, and professional applications. Understanding how to effectively define and invoke personas opens up a vast landscape of possibilities for interacting with these advanced AI models.