Sherlockian Prompt Design(1)
Sherlockian Way of Thinking
1. Introduction: Observe, Organize, and Ask
A prompt is not merely a sentence. It is the shape of thought, the architecture of a question, the expression of intent. In the previous chapter, through Sherlock Holmes’s six methods of reasoning, we learned that a prompt is more than a request: it is a logically assembled structure of thinking.
Yet there is a clear gap between “knowing” a logical structure and “designing” a question that actually uses it. It is the difference between reading about Holmes’s methods and standing in a crime scene, arranging the clues with your own hands.
This chapter begins the practical training to bridge that gap. For each reasoning method, we will stage real-world problem scenarios and practice how to construct prompts that fit them. We will then go further, combining two or three methods so that you can tackle more intricate problems.
What matters is not a perfect sentence, but a design of thought that elicits a meaningful response from an AI. In the process, you will not only handle AI more deftly but also strengthen your own power to structure and express thinking.
Holmes did not pontificate. He observed, organized, and asked. That is exactly what we will do now: structure our thoughts and translate that structure into the form of a question. It is time to bring the methods you have learned down into your fingertips.
☞ How to use this chapter
For each method, read the “Scenario” carefully. Under “Your Prompt,” draft the best prompt you can. At the end of the chapter, compare with the sample prompts and design notes, and then revise your prompt once more.
2. Deduction: Narrow Answer by Stated Conditions
Deduction applies a general principle to a specific case to reach a conclusion.
Using concrete situations, let us practice how to state conditions and structure the question.
Exercise ① Getting Book Recommendations That Fit Your Taste
Scenario. Lately you struggle to choose what to read. You prefer clear, information-forward books over moving stories, want something you can finish quickly, and—above all—seek knowledge you can use in intelligent conversation.
Your Prompt:
(Write here.)
Exercise ② Designing an Education Content Module
Scenario. You are designing educational content about AI ethics for teenagers. Your goal is to make the abstract topic of “AI ethics” understandable and engaging for a 15-year-old audience. The following guidelines should be followed:
Each lesson should fit within a 40-minute class period.
Focus on compelling real-world cases rather than dry theory.
Include at least one discussion topic per session.
Use actual case studies based on domestic or international AI-related events and issues.
Your Prompt:
(Write here.)
Closing note: The aesthetics of constraint
Many believe prompts work best when written “freely.” In truth, AI prefers clarity to freedom. Deductive prompts do not presume omniscience; they design a frame of conditions and narrow the answer within it. A good prompt does not display infinite possibilities—
(continued in samples below).
2. Deduction — Sample Section
Exercise ① Book-taste alignment
Sample Prompt.
I'm looking for books that are more information-driven than narrative and under 200 pages in length. They should cover topics like current affairs, society, or science, and offer insights that can be useful in thoughtful, intellectual conversations. Recommend just three books, and explain in 2–3 sentences why each one is worth reading.
Design note.
Instead of a single keyword, structure constraints as you would in deduction to intentionally narrow the scope. The quality and validity of recommendations rise accordingly.
Exercise ② Education content design
Sample Prompt.
“Please design a 40-minute AI ethics lesson for middle-school students. Use real domestic and international cases to drive comprehension, and include discussion prompts for students to share their views. Friendly, accessible language is preferred.”
Design note.
This is not “recommend content,” but “design to specified conditions.” You are assigning the role of educator to the AI and asking it to meet a constraint set.
