AI language models like ChatGPT and Claude are growing more sophisticated by the day, literally. For now, prompting is how to get the best out of them, and, as usual, English-based ideas are dominating. But also, as usual, Japan is doing its own innovation. And you probably don’t know about it.

Being a foreign-owned company in Japan, we hope to convey Japan’s breakthroughs every now and then. That’s what this article is for.

Two Japanese innovators, Takayuki Fukatsu and Shunsuke Hayashi, have developed unique prompting methods that, like Spinal Tap (boomer reference), are “big in Japan,” so let’s explore these techniques and how they’re shaping the future of AI interactions.

Table of Contents

Fukatsu (Resurrection method) prompting

Definition of Fukatsu prompting

The Fukatsu-style prompt, also known as the resurrection-style prompt, is a structured approach for interacting with ChatGPT and other gen-AI chatbots.

This method, developed by Takayuki Fukatsu, who works on GPT chat systems, aims to provide the AI with clear and specific instructions.

Here’s a summary of the steps:

Clear structure
Four main elements here:
1. Instruction (命令 meirei)
2. Constraints (制約 seiyaku)
3. Input (入力 nyuryoku)
4. Output (出力 shutsuryoku)
Define the AI’s role: Clearly state the role the AI should assume, such as ‘You’re an expert content strategist”.
Give specific instructions: Use clear and detailed guidance on what you want the AI to do.
Set constraints: Give specific conditions, like word count, tone, and target audience.
Input information: Provide the necessary data or context for the AI to work with.
Specify output format: Clearly define what response format you want, such as bullet-pointed or in paragraphs.
Use markup language: Employ markup to clearly distinguish between the prompt’s different elements (though this isn’t mandatory; it just keeps things cleaner).
Use bullet points or similar: Use a list to present the instructions and constraints clearly and concisely.
Have the AI to ask questions: End the prompt with “If you need additional information to achieve the best results for this task, ask me questions”.

This structure lets you create effective prompts that lead to higher-quality, more targeted responses.

You may be thinking that that’s a lot to write, but I’d encourage you to weigh it against the need for long iterations used in other prompting methods. You’re preempting the need for so many revisions and, in my case, occasional frustration.

Example of Fukatsu prompting

In markup, the Fukatsu prompt goes like this:

#Command
You are (role).
Please output the best (topic of the article) based on the constraints and input statements below.

#Constraints
(explain constraints)

#Input
(question or instruct)

#Output

So here’s how you might structure a very basic Fukatsu-style prompt with the four main elements.

Command:
You’re a professional editor. Provide a sharp and conversational summary of the following text.

Constraints:
The summary should be around 300 characters, easily understood by elementary school students. Make the sentences concise and don’t omit any essential keywords.

Input:
You’ll put your target text here.

Output: [the AI will provide it]

This structure guides the AI to produce a summary that meets specific criteria, making it easier to get results that deliver what you’re after.

Fukatsu compared with Shunsuke prompting

While both Fukatsu and Shunsuke methods (described below) aim to improve AI interactions, they differ in their approach and focus. The Fukatsu method emphasizes clear instructions and role definition, making it well-suited for tasks that require specific outputs. However, the Shunsuke (Goal-Seek Prompt) method, focuses on guiding the AI to ask questions and clarify the user’s intent.

The Fukatsu method typically results in more direct responses because it provides a clear framework for the AI to follow. The Shunsuke method often leads to a more interactive exchange, with the AI asking follow-up questions to better understand the user’s goals.

For tasks that require precise outputs, such as data analysis or content creation with specific parameters, the Fukatsu method might be more effective. However, the Shunsuke method could yield better results for more open-ended or creative tasks where the user’s goals aren’t fully formed.

How Fukatsu compares with other prompting techniques

Like most prompting methods, Fukatsu has some similarities with other methods and some unique twists. And these methods are constantly being iterated in this very dynamic space.

For instance, the RISEN framework (Role, Instructions, Steps, End goal, Narrowing) also emphasizes role definition and clear instructions, while the Fukatsu method’s structure is generally simpler and more focused on the output specifications. In “few-shot prompting,” you give examples of desired outputs within the prompt. This can be incorporated into a Fukatsu-style prompt, but it’s not a core component of this method.

Fukatsu prioritizes clear instructions over examples, making it potentially more versatile across different task types. In comparison, “chain-of-thought” prompting encourages the AI to break down its reasoning process. While this isn’t necessarily part of the Fukatsu method, users can incorporate this approach within a Fukatsu prompt’s constraints or command sections to see the AI’s reasoning.

Overall, the Fukatsu method gives you a good balance of structure and flexibility, so you can guide the AI responses without using overly complex prompts.

Pros and cons of Fukatsu prompting

Pros

Clarity: The structured approach reduces ambiguity in the AI’s responses.
Consistency: You can maintain a uniform approach across different interactions.
Versatility: Works well for a wide range of tasks and domains.
Efficiency: Clear instructions often lead to more accurate first-attempt responses.

Cons

Learning curve: It can take some time to master this structure.
Potential for over-specificity: Extremely detailed prompts could limit AI creativity.
Length: Comprehensive Fukatsu prompts run longer than simpler queries.
Model dependency: The effectiveness may vary across different AI models.

Shunsuke (Goal-Seek Prompt) prompting

Definition of Shunsuke prompting

The Shunsuke method (Goal-Seek Prompt method) was developed by Shunsuke Hayashi at XinobiAI. This approach focuses on guiding the AI to ask questions and help users clarify their objectives.

This method, unlike more directive and unilateral methods, tries to build a collaborative dialogue between the user and the AI. The core principle of this method is to transform vague or ambiguous user inputs into concrete, actionable goals. To do so, you instruct the AI to engage in a series of clarifying questions that help to refine your intent and desired outcome.

Example of Shunsuke prompting

A Shunsuke-style prompt in markup is structured like this:

#Content detail
This content is (topic).

#Variables
[var1] = val1
[var2] = val2
[var3] = val3

#Command
Please write about (first requested task) based on [var2] for [C1] = [var1].
Please write about (second requested task) based on [C2] = [C1].

#Execute
$ run [C1] [C2]

It’d go something like this:

1. Initial instruction: “You’re an AI assistant trained to help users clarify their goals. Your task is to ask questions that will help refine the user’s objective.”

2. User input: “I want to improve my writing.”

3. AI response (following the Shunsuke method): “I’d be glad to help you improve your writing. To better understand your goal, could you please answer these questions: o What type of writing do you want to improve? (e.g., academic, creative, business)

Are there specific aspects of writing you want to focus on? (e.g., grammar, style, structure)
What’s your current skill level in writing?
Do you have any specific projects or deadlines in mind?”

This example shows how the Shunsuke method guides the AI to ask probing questions, helping to clarify the user’s goals and provide more targeted assistance.

Shunsuke compared with Fukatsu prompting

The Shunsuke method differs from Fukatsu prompting in several ways:

Interaction style: Shunsuke prompting encourages a back-and-forth dialogue, while Fukatsu prompting aims for a more direct, single-response output.
Goal definition: The Shunsuke approach helps users refine their goals through questioning, while Fukatsu assumes the goal is clear from the outset.
Flexibility: Shunsuke prompting adapts to user responses, making it suitable for open-ended tasks. Fukatsu prompting is good in scenarios when the outcome you want is well-defined.
Structure: Fukatsu prompts have a more rigid structure with specific components, while Shunsuke prompts are more fluid and adaptable.

How Shunsuke compares with other prompting techniques

The Shunsuke method shares similarities with other AI prompting techniques:

Socratic questioning: Both approaches use targeted questions to deepen understanding, but Shunsuke is specifically designed for AI-human interactions in goal clarification.
Iterative refinement: Like some advanced prompting techniques, the Shunsuke method includes multiple rounds of interaction. What’s different is its focus on goal clarification rather than output refinement.
Conversational AI: Many chatbots play well with conversational techniques, but the Shunsuke method is more structured in its approach to goal-seeking.
Coaching frameworks: The Shunsuke method is similar to coaching techniques that help clients clarify their objectives, but adapted for AI–human interactions.

Pros and cons of Shunsuke prompting

Pros:

Goal clarity: Helps you refine vague ideas into concrete objectives.
Adaptability: Works well for a wide range of topics and user skill levels.
User engagement: Encourages your active participation.
Learning opportunity: You’ll often gain insights through the questioning process.

Cons:

Time-consuming: The back-and-forth nature can take longer than direct prompting.
Frustration potential: You might prefer more immediate answers.
Dependency on user input: The outcome quality relies heavily on your ability to articulate your thoughts.
Complexity: Implementing this method effectively requires careful design of the AI’s questioning strategy.

The Shunsuke (Goal-Seek Prompt) method, while valuable, isn’t the best choice for every situation. For quick, straightforward tasks, more direct prompting methods are probably better. But the Shunsuke method is a strong choice for complex problems or when you’re not exactly sure what you’re after.

As AI technology continues to evolve, we can expect to see further refinements and new applications of these two methods. I’m always confident in Japan’s tech minds to humbly come up with solutions, but less confident in their own confidence in showing them off. I’m here to help in any way I can. I’m fascinated with how conversational AI essentially removes the language barrier.

There’s no technical or linguistic excuse for foreign-language-based ideas not to be rapidly translated and communicated. I’ll keep working with the tools available to dig for these gems and share them.

Here at MacroLingo, we bring you expert-driven human-generated content. And these days, that means leveraging generative AI and the latest tools. When you want to write about gen AI, learn to use it in your company, or you’re advancing new technology, we have a full range of content solutions, including writing, editing, and SEO/GEO (generative search). Get in touch with any ideas for how we can collaborate or make your business more profitable and future-proof.