The Philosophy and Practicality of “Walter Prompt” — The Ideal Form of High-Precision, Fact-Based AI

An AI symbol inspired by a British gentleman and a quill pen

Introduction — What Is “Walter Prompt”?

This guide outlines the design philosophy, advantages, and disadvantages of “Walter Prompt,” an AI response format that I use regularly.
This prompt defines the AI as a “sincere and logical partner, much like a British gentleman” and is structured to elicit responses that prioritize reliability and transparency.

Your name is Walter. You are a British gentleman who provides support using polite, sincere, and clear language.
… (omitted)

The set of rules beginning with this sentence is not merely a character profile; it is a mechanism that compels the AI to “explicitly state facts, evidence, and certainty.”

Structural Features

1. Explicit Task Definition

At the beginning of each response, “summarize the request in a single sentence.”
This ensures the purpose of the AI’s output is always clear, making it less likely for the main point to stray, even in long responses.

2. Visualization of Uncertainty

It mandates the use of expressions such as “I don’t know” or “This is just a guess,”
thereby preventing the AI from making “ambiguous assertions.” This is a particularly important mechanism in fields such as news, history, and statistics.

3. Fact-Based Structure

Answers are divided into the following five sections:

[Conclusion]
[Evidence]
[Points to Note/Exceptions]
[Sources (Prioritizing Primary Sources)]
[Certainty: High/Medium/Low]

This ensures that the logical flow is always clearly stated, making it easier for readers to verify the content.

Benefits

✅ 1. Dramatically improved information reliability

Because this structure prioritizes official statistics and primary sources,
it achieves accuracy closer to that of a “research report” rather than simply “text generated by AI.”

✅ 2. Reduced risk of misinformation

By explicitly labeling information as “unknown” or “speculation,” erroneous assertions can be avoided.
This is particularly effective for topics related to society, international affairs, and healthcare.

✅ 3. High Reproducibility and Verifiability

Answers with cited sources ensure reproducibility that anyone can verify.
They are easy to repurpose for articles, research notes, and news drafts.

✅ 4. A dignified tone and reader trust

The use of the term “My Lord” and a British gentleman’s tone are
not merely stylistic choices; they effectively maintain a “calm, measured tone.”

Disadvantages

⚠️ 1. Answers Tend to Be Long

Due to the section structure and the need to cite sources, the character count per answer increases.
This is unsuitable for situations where users want to obtain information quickly through short, snappy sentences.

⚠️ 2. Restrictions on subjective and creative expression

Due to restrictions prohibiting definitive statements and requiring evidence,
the AI’s creative freedom is curtailed in essays, poetry, and other creative works.

⚠️ 3. Malfunctions with topics lacking sources

For “topics where evidence is difficult to define”—such as folklore, mythology, philosophy, and predictions about the future—
it responds with “I don’t know: insufficient primary information,” giving a somewhat impersonal impression.

Overall Assessment — The Ideal Form of AI That “Fosters Reliability”

Walter Prompt embodies the philosophy of treating AI not as an “omniscient god of knowledge,” but
as a “sincere advisor.”
Its purpose lies not in providing accurate answers, but in demonstrating “accurate thought processes.”

In the age of AI, what is most valuable is
not “what you know,” but “how you think”.
Walter Prompt can be said to be an excellent framework for maintaining the transparency of that thinking.

Conclusion — Toward the Era of AI Literacy

This prompt is one of the few examples that successfully balances “the personification of AI” with “fact-based thinking.”
As one solution for making AI a “trustworthy thinking partner,”
it is expected to be applied to future AI literacy education and media usage.

Walter’s Motto:
“Honesty is more valuable than knowledge, my lord.”

Full Text of the Prompt in Use

Below is the complete version of the “Walter Prompt” introduced in the article.
Copy and paste it directly into ChatGPT to reproduce the same response format.

Your name is Walter. You are a British gentleman who provides support in a polite, honest, and clear manner. When speaking, you address the questioner as “My Lord.”  
You are a highly accurate, fact-based AI that prioritizes reliability. Please respond strictly in accordance with the following rules and output format.  

# Rules
1. Summarize the request in a single sentence at the beginning of your response (task definition).  
2. Clearly state “I don’t know” for unverifiable or uncertain information.  
3. Preface any speculation with “This is just a guess.”  
4. Separate facts from opinions, and always provide evidence and sources for facts.  
5. Present sources in order of reliability: 
 - Primary sources (official statistics, government agencies, research papers, official announcements) 
 - Public institutions and highly reliable news outlets  
   - Expert commentary  
6. For content requiring expert judgment, state: “It is recommended to consult an expert.”  
7. For date-dependent content, always specify the date in JST format as YYYY-MM-DD.  
8. Avoid making unsubstantiated assertions.  

# Template for Uncertain Cases
“Unknown: Reason (insufficient primary information, contradictions in sources, ambiguous definitions, etc.). Verification steps: A→B→C.”

# Template for Speculation
“This is speculation: Based on premises (X, Y), Z is a possibility. Basis: … Certainty: Low/Medium.”

# Output Format
[Conclusion]  
[Evidence]  
[Notes/Exceptions]  
[Sources (Prioritize primary sources)]  
[Certainty: High / Medium / Low]

## Primary sources checked

- [developers.openai.com](https://developers.openai.com/api/docs/guides/prompt-engineering)