- Published on
What Socratic Maieutics and Prompt Engineering Have in Common
"Teacher, how do I ask ChatGPT questions to get the answers I want?" A student asked me this during class, and I paused. And then it came to me. Socrates. Two thousand four hundred years ago, he did not pour knowledge into his students. Instead, he used questions to awaken the knowing that slept within them. The way we converse with AI today β prompt engineering β is remarkably similar to that maieutics. The truth that good questions produce good answers transcends time.
Table of Contents
- What Is Maieutics: Socrates' Methodology
- The Essence of Prompt Engineering
- The Structural Similarities Between the Two Methods
- Socratic AI Use in the Classroom
- The Questioning Human: A Core Competency for the AI Age
1. What Is Maieutics: Socrates' Methodology
The Role of the Midwife
Socrates' mother was a midwife. He described himself as a kind of midwife too β not delivering babies of the body, but drawing out the knowing that sleeps within the soul. This is "maieutike," the art of midwifery.
Meno's Paradox and the Theory of Recollection
In Plato's dialogue Meno, Socrates teaches a slave boy with no knowledge of geometry the Pythagorean theorem. The remarkable thing is that he never once gives the answer. Using only questions, he guides the boy to arrive at the answer himself. Socrates called this "anamnesis" β recollection: our souls already know, and learning is the act of remembering.
Characteristics of Socratic Questions
- Open-ended questions: ones that cannot be answered with yes or no
- Questions that expose assumptions: "What is your reason for thinking that is self-evident?"
- Questions that confront contradictions: "How do those two things you just said coexist?"
- Questions that dig deeper: "Then what is this 'goodness' you speak of?"
2. The Essence of Prompt Engineering
What It Means to Ask AI Good Questions
"What answer AI produces depends entirely on how you ask." This is the core proposition of prompt engineering. Asking the same AI "write me a marketing plan" versus "You are a B2B SaaS marketing specialist with ten years of experience. Propose a step-by-step customer acquisition strategy for a startup's first six months" produces completely different results.
Core Elements of Prompt Design
- Role setting: Assigning AI the persona of a specific expert
- Context provision: Richly providing situational and background information
- Constraints: Specifying the desired format, length, and perspective
- Iterative refinement: Using the first answer as the basis for follow-up questions
Chain-of-Thought: Questions That Guide the Thinking Process
The instruction "think through this step by step before giving your answer" dramatically improves the quality of AI reasoning. This is no different from Socrates saying to a student, "Don't tell me the conclusion first β let's explore the process of thinking that led you there together."
3. The Structural Similarities Between the Two Methods
Similarity 1: Process Over Answer
Neither Socrates nor a good prompt simply demands a conclusion. Both guide the unfolding of the process of thinking. Socrates never directly defined "justice." A good prompter does not ask "what is the conclusion?" but rather "analyze what steps need to be taken to solve this problem first."
Similarity 2: Dismantling Assumptions
Socrates constantly shook the premises his conversation partners took for granted. "The thing you think you know β do you really know it?" A good prompt does the same. Asking AI "point out any assumptions or biases I may be making in this analysis" is performing Socratic self-reflection together with AI.
Similarity 3: Awareness of Ignorance
Socrates' famous declaration β "I know that I know nothing." This is where knowing begins. In prompt engineering too, the awareness that "I'm not sure exactly what I want" leads to more precise questions. Vague requests produce vague answers.
Similarity 4: Iterative Dialogue
Maieutics does not end with one question. It involves continuous conversation and deepening inquiry. Prompt engineering is the same. The first answer is just the starting point. "Be more specific," "from a different angle," "give me a counterexample" β these follow-up questions draw out genuine insight.
4. Socratic AI Use in the Classroom
Teaching Prompts Means Teaching Questions
It is not simply about teaching how to use AI. It is about developing the capacity to design questions. This is actually the same thing Socrates was trying to teach β the ability to think clearly and express your thinking in precise language.
Classroom Application Examples
In Korean language or social studies classes:
"Before asking AI about this historical event, let's first make a list of what we want to know. Which questions will lead to deeper understanding?"
Critical thinking training:
"Don't just copy down AI's answer. Ask AI again: 'What are the weaknesses of this argument?' Then examine that answer too."
Creative writing:
"Before throwing the topic at AI, ask yourself like Socrates would: what do I actually want to know about this topic?"
The Paradox of Socratic AI Use
When AI is used in a Socratic way, the one who learns the most is not the AI but the student. In order to formulate a good question, the student must examine their own thinking, confront what they do not know, and think more clearly. AI becomes a mirror.
5. The Questioning Human: A Core Competency for the AI Age
From the Age of Answers to the Age of Questions
When information was scarce, the person with good answers was precious. In an age when AI produces all the answers, the person who can pose good questions is precious. Paradoxically, the advance of AI has summoned Socrates once more.
Redefining Education
What should we teach students?
- Memorizing information β AI replaces this
- Searching for information β AI replaces this
- Asking the right questions of information β human competency
- Interpreting the meaning of answers β human competency
- Integrating knowledge toward a better life β human competency
A Final Reflection
Socrates never stopped his inquiry, even as he drank the hemlock. His legacy was not specific knowledge but the disposition to question itself. What we can pass on to students in the AI era is no different: the courage to keep asking, the honesty to face what we do not know, and the intellectual humility to press toward better questions.
How do you design your questions when conversing with AI? If you have experienced asking a question that shook an assumption, share it in the comments.
Related Reading