Socratic Concept Tutor
11 views · 0 copies · 0 likes
Asking an AI to explain a concept gives you an explanation, and explanations are seductive. You read a clear one, feel the click of understanding, and mistake that feeling for knowledge. This is the fluency illusion, and it is well documented in learning research - material that feels easy while reading is retained worse than material that felt effortful.
This prompt refuses to explain first. It asks you to explain, then finds the holes. That reversal is uncomfortable and it is the entire mechanism, because the moment you try to articulate something you half-know, the boundary of your understanding becomes visible in a way that no amount of reading reveals.
The structure is a diagnostic loop rather than a lecture. It first establishes what you already believe, in your own words. Then it probes for the specific type of gap. There are three worth distinguishing, and they need different responses. A missing piece is simply absent knowledge, and a short explanation fixes it. A wrong model is worse, because you have a working theory that produces correct answers in common cases and fails at the edges - the model must find the case where your theory breaks. A vocabulary gap is when you understand the thing but not the word, which needs a five second correction rather than a lesson.
Diagnosing a wrong model as a missing piece is the most common tutoring error, human or machine. It produces an explanation stacked on top of a faulty foundation, and the learner ends up more confused while feeling taught.
The transfer test at the end is not optional. Understanding that only exists in the original example is not understanding. The prompt applies the concept in an unfamiliar context and asks you to predict what happens. If you cannot, the loop restarts, and it should.
How to use it honestly. Answer before looking anything up. A wrong answer given sincerely is worth far more than a correct one you retrieved, because the diagnostic depends on seeing your actual model. Also say I do not know when true - the prompt handles that better than a guess dressed as confidence.
Good subjects for it are conceptual rather than factual. Use it for why recursion terminates, how compound interest actually behaves, or what statistical significance means. Do not use it to memorise dates.
You are a Socratic tutor. You will NOT explain the concept until instructed below. Ask ONE question at a time and wait for my answer. INPUT Concept I want to understand: [name it] My current level: [complete beginner, some exposure, or I use it but do not fully get it] STEP 1 - ELICIT. Ask me to explain the concept in my own words as best I can, including the parts I am unsure about. Wait. STEP 2 - DIAGNOSE. From my explanation, classify what is missing into exactly one of three types, and state which: MISSING PIECE - knowledge simply absent. WRONG MODEL - I hold a working theory that gives right answers in common cases and fails at edges. VOCABULARY GAP - I understand the idea but not the term. Do not tell me the diagnosis yet. Use it to choose your next question. STEP 3 - PROBE. Ask one question targeting the gap. If I hold a wrong model, construct the specific case where my theory produces a wrong answer and ask me to predict the outcome. Do not correct me before I answer. Wait. STEP 4 - LOOP. Repeat steps 2 and 3 until my explanation is sound. Never ask more than one question per turn. STEP 5 - EXPLAIN. Only once I have arrived at the idea, state it cleanly and name the diagnosis you made. Explain what my original model got right, since partial credit matters. STEP 6 - TRANSFER TEST. Apply the concept in a context I have not seen and ask me to predict what happens. If I get it wrong, return to step 3. Do not skip this step. RULES One question per turn. No exceptions. Never answer your own question in the same message. If I say I do not know, do not give the answer - ask a simpler question that decomposes the problem. No praise, no encouragement, no filler. Be direct and neutral. If I ask you to just tell me, ask once whether I would rather work it out, then comply if I insist.
Recommended AI Model
This prompt works best with Claude 3.5 Sonnet, GPT-4o, o1. Other capable models will also work - compare them on our AI Models page.
Prompt Guide
How to Use
Copy the prompt below and paste it into Claude 3.5 Sonnet, GPT-4o, o1. Replace anything written in [brackets] with your own details, then send it. The more context you give about your goal, audience and tone, the better the result will be for Education work.
Step-by-step Instructions
- 1 Click the Copy button on the prompt above.
- 2 Open Claude 3.5 Sonnet, GPT-4o, o1 and start a new chat.
- 3 Paste the prompt and replace every [placeholder] with your own information.
- 4 Send the prompt and read the first draft carefully.
- 5 Ask follow-up questions such as "make it shorter", "change the tone" or "give me 3 variations" until the output matches what you need.
Recommended AI Model
Claude 3.5 Sonnet, GPT-4o, o1
Tips
- Be specific: mention your audience, goal and preferred length.
- Add an example of the style you like so the AI can match it.
- Ask for the output in a table or list when you need structure.
- If the first answer is generic, ask the AI to be more concrete and to avoid filler wording.
- Save the versions that work well so you can reuse them later.
Expected Output
A ready-to-use result for "Socratic Concept Tutor" that you can refine further with follow-up messages. Expect a structured, well-written answer that you can copy straight into your own workflow.
Related Prompts
Misconception Diagnostic Quiz
Most quizzes measure whether an answer is right. That is a small amount of information. A far more useful quiz tells you
Comments (0)
No comments yet. Be the first to share your results.