TL;DR

• This is a creativity / interdisciplinary test: combine seemingly unrelated concepts and see if the model can generate coherent, style-consistent, and "novel" text.
• Good for: ad creative, brand copy, story/character development, and "cross-domain analogies."
• Production tip: specify style, length, structure (paragraphs/bullet points), and banned items (avoid offense, avoid fabricating sensitive info about real people).

Background

The following prompt tests an LLM's capabilities to perform interdisciplinary tasks and showcase its ability to generate creative and novel text.

How to Apply

Think of this prompt as a "style transfer + concept mashup" template:

• First, pick the narrator/voice (e.g., Mahatma Gandhi's letter-writing style)
• Then define the target and conflict (Electron running as a US presidential candidate)
• Finally, add constraints: what points must be included, output length, whether to use quotes/metaphors

How to Iterate

1. Add structural requirements: Opening / Arguments / Counterarguments / Closing
2. Add style constraints: vocabulary, rhythm, rhetorical devices (metaphor / parallelism)
3. Add safety constraints: avoid improper attribution to real people; use fictional personas if needed
4. Add self-check: after output, list 3 creative highlights + 1 weakest point

Self-check Rubric

• Is the text coherent and logically consistent?
• Is the style consistent (does the voice stay in character)?
• Are the interdisciplinary elements genuinely fused, not awkwardly stitched together?
• Is the structure reusable (can you swap the topic and keep going)?

Practice

Exercise: replace the prompt elements with things from your industry:

• Use a historical figure's voice to write a "supporting letter"
• Turn a scientific concept/tech term into a candidate/product
• Require output with 3 arguments + 1 counterargument response

Prompt

Write a supporting letter to Kasturba Gandhi for Electron, a subatomic particle as a US presidential candidate by Mahatma Gandhi.

Code / API

OpenAI (Python)

from openai import OpenAI

client = OpenAI()

response = client.chat.completions.create(
    model="gpt-4",
    messages=[
        {
            "role": "user",
            "content": "Write a supporting letter to Kasturba Gandhi for Electron, a subatomic particle as a US presidential candidate by Mahatma Gandhi.",
        }
    ],
    temperature=1,
    max_tokens=1000,
    top_p=1,
    frequency_penalty=0,
    presence_penalty=0,
)

Fireworks (Python)

import fireworks.client

fireworks.client.api_key = "<FIREWORKS_API_KEY>"

completion = fireworks.client.ChatCompletion.create(
    model="accounts/fireworks/models/mixtral-8x7b-instruct",
    messages=[
        {
            "role": "user",
            "content": "Write a supporting letter to Kasturba Gandhi for Electron, a subatomic particle as a US presidential candidate by Mahatma Gandhi.",
        }
    ],
    stop=["<|im_start|>", "<|im_end|>", "<|endoftext|>"],
    stream=True,
    n=1,
    top_p=1,
    top_k=40,
    presence_penalty=0,
    frequency_penalty=0,
    prompt_truncate_len=1024,
    context_length_exceeded_behavior="truncate",
    temperature=0.9,
    max_tokens=4000,
)

Reference

• Sparks of Artificial General Intelligence: Early experiments with GPT-4 (13 April 2023)