​
What is Socratic Questioning?

Socratic questioning is a prompting technique based on the Socratic method of inquiry, which uses systematic questioning to explore ideas, uncover assumptions, analyze concepts, and guide critical thinking. In the context of AI, it involves asking a series of thoughtful, open-ended questions that lead the AI through a structured reasoning process rather than directly asking for answers.

​
Why Use Socratic Questioning?

  • Deeper Understanding: Encourages exploration of underlying concepts and assumptions
  • Improved Reasoning: Guides the AI through organized, logical thinking processes
  • Reduced Bias: Questions help uncover and examine hidden assumptions and biases
  • Better Problem Solving: Breaks complex problems into manageable components
  • Enhanced Critical Thinking: Promotes evaluation of multiple perspectives and alternatives
  • Guided Discovery: Allows the AI to reach conclusions through guided inquiry rather than direct instruction

​
Basic Implementation in Latitude

Here’s a simple Socratic questioning example for exploring a concept:
Concept Exploration
---
provider: OpenAI
model: gpt-4o
temperature: 0.5
---

# Socratic Exploration: {{ concept }}

Let me guide our exploration of this concept through systematic questioning:

## Initial Definition:
What is meant by {{ concept }}? How would you define it in simple terms?

## Core Assumptions:
What assumptions underlie this concept? What must be true for this concept to make sense?

## Evidence Examination:
What evidence supports this concept? What evidence might challenge it?

## Implications:
What follows if this concept is valid? What are the logical consequences?

## Alternative Perspectives:
How might someone with a different viewpoint understand this concept?

## Synthesis:
Based on our examination, how might we now understand {{ concept }} more deeply?

​
Advanced Implementation with Directed Reasoning

Let’s create a more sophisticated example that uses Socratic questioning to guide complex problem-solving: 
---
provider: OpenAI
model: gpt-4o
temperature: 0.6
---

<step>
# Socratic Problem Analysis

Let's explore this problem through careful questioning:

## The Problem:
{{ problem_statement }}

## Understanding the Problem:
1. What are the key elements of this problem?
2. What are we trying to determine or accomplish?
3. What information is provided, and what information is missing?
4. How will we know when we've reached a solution?

## Initial Thoughts:
Let me consider what approaches might be relevant to this problem...
</step>

<step>
# Examining Assumptions

Let's question the assumptions underlying our approach:

## Identified Assumptions:
1. What assumptions am I making about this problem?
2. Are there unstated assumptions in the problem statement itself?
3. Which of these assumptions might be challenged?
4. How would changing these assumptions affect our approach?

## Refining Our Understanding:
Based on examining these assumptions, I should revise my understanding as follows...
</step>

<step>
# Exploring Alternative Approaches

Let's question our approach and consider alternatives:

## Current Approach:
The approach I'm considering is...

## Questioning the Approach:
1. Why might this approach work for this problem?
2. What are the weaknesses or limitations of this approach?
3. What alternative approaches could be considered?
4. How do these approaches compare in terms of effectiveness and efficiency?

## Selected Approach:
After this examination, the most promising approach appears to be...
</step>

<step>
# Working Through the Solution

Let me use Socratic questioning to work through the solution:

## Solution Path:
1. What is the first step in solving this problem?
2. How does this step lead to the next?
3. What obstacles might arise, and how can they be addressed?
4. How can I verify that each step is correct?

## Solution Development:
Working through these questions systematically...
</step>

# Evaluating the Solution

Let's critically examine our solution:

## Solution Review:
1. Does this solution fully address the original problem?
2. What are the strengths and weaknesses of this solution?
3. How could this solution be improved?
4. Under what conditions might this solution fail?

## Final Solution:
Based on this evaluation, my revised solution is:

**Answer:**
**Confidence:**
**Key Insights:**
In this advanced example:
  1. Progressive Inquiry: Questions build systematically from problem understanding to solution evaluation
  2. Chain Processing: Each step deepens the inquiry through targeted questioning
  3. Self-Reflection: The AI questions its own assumptions and approaches
  4. Critical Evaluation: The solution undergoes rigorous examination before finalization

​
Ethical Reasoning with Socratic Questioning

Use Socratic questioning to explore ethical dilemmas: 
---
provider: OpenAI
model: gpt-4o
temperature: 0.6
---

<step>
# Ethical Dilemma Analysis

Let's explore this ethical situation through Socratic questioning:

## The Situation:
{{ ethical_dilemma }}

## Understanding the Dilemma:
1. What are the key ethical considerations in this situation?
2. Who are the stakeholders affected by this situation?
3. What values or principles are potentially in conflict?
4. What information would help us better understand the ethical implications?
</step>

<step>
# Examining Perspectives

Let's question from multiple ethical frameworks:

## Consequentialist Perspective:
1. What are the potential outcomes of each possible action?
2. Who benefits and who is harmed by each outcome?
3. How do we weigh different types of benefits and harms?
4. What unintended consequences might arise?

## Deontological Perspective:
1. What duties or obligations apply in this situation?
2. Are there any absolute rules that should not be broken?
3. What rights do the various parties have?
4. What would happen if everyone acted this way?

## Virtue Ethics Perspective:
1. What would a virtuous person do in this situation?
2. What character traits are being developed or diminished?
3. How do intentions factor into the ethical evaluation?
4. What does this situation reveal about our values?
</step>

# Synthesizing Ethical Insights

Let's bring together our ethical exploration:

## Areas of Agreement:
What common ground exists between different ethical perspectives?

## Key Tensions:
What fundamental tensions remain unresolved?

## Contextual Factors:
How do specific contextual factors influence the ethical evaluation?

## Ethical Conclusion:
After this systematic questioning, my ethical assessment is:

**Recommended Action:**
**Ethical Justification:**
**Remaining Concerns:**
**Conditions for Reassessment:**

​
Multi-Agent Socratic Dialogue

Combine Socratic questioning with multiple perspectives for deeper exploration: 
---
provider: OpenAI
model: gpt-4o
temperature: 0.7
type: agent
agents:
  - agents/questioner
  - agents/responder
  - agents/critic
---

# Socratic Dialogue on: {{ topic }}

This dialogue will explore {{ topic }} through a structured Socratic conversation between multiple agents.

## Initial Problem Statement:
{{ problem_statement }}

## Dialogue Process:
1. **Questioner**: Ask probing questions to explore the topic
2. **Responder**: Provide tentative answers and explanations
3. **Critic**: Evaluate the reasoning and suggest improvements
4. **All**: Work together to reach deeper understanding

Begin the Socratic dialogue to explore this topic thoroughly.

​
Best Practices for Socratic Questioning

Effective Questioning:
  • Start with broad, open-ended questions before narrowing focus
  • Use clear, concise, and unambiguous language
  • Ask one question at a time rather than bundling multiple questions
  • Balance questions that explore breadth with those that explore depth
  • Use “what,” “how,” and “why” questions to encourage elaboration
  • Avoid leading questions that suggest a particular answer
Question Progression:
  • Begin with conceptual clarity questions before moving to analysis
  • Progress from factual to interpretive to evaluative questions
  • Use follow-up questions that respond directly to previous answers
  • Include metacognitive questions that reflect on the thinking process itself
Productive Dialogue:
  • Allow appropriate “thinking time” between questions
  • Acknowledge and build upon previous responses
  • Maintain a tone of genuine curiosity rather than interrogation
  • Balance challenging questions with supportive responses
  • Recognize when to shift direction if a line of questioning becomes unproductive
  • Periodically summarize insights before moving to new areas
Common Pitfalls:
  • Asking too many questions without allowing for deep exploration
  • Pursuing trivial lines of inquiry that don’t advance understanding
  • Forcing a predetermined conclusion rather than following the inquiry where it leads
  • Failing to adapt questions based on previous responses
Best Use Cases:
  • Complex ethical dilemmas
  • Concept exploration and clarification
  • Critical analysis of arguments and claims
  • Problem-solving requiring deep consideration
  • Examining assumptions and biases
  • Teaching and educational contexts
Less Suitable Cases:
  • Simple factual questions with clear answers
  • Emergency situations requiring immediate decisions
  • Highly specialized technical problems
  • Tasks focused on creative generation rather than analysis
Six Types of Socratic Questions:
  • Conceptual clarification: “What exactly do you mean by…?”
  • Probing assumptions: “What are you assuming when you say…?”
  • Probing reasons and evidence: “What evidence supports that view?”
  • Questioning viewpoints and perspectives: “How might others see this differently?”
  • Exploring implications and consequences: “What follows from that position?”
  • Questions about the question: “Why is this question important?”
Strategic Application:
  • Match question types to specific reasoning needs
  • Use different question types in a complementary sequence
  • Adapt question depth based on the complexity of the topic
  • Consider using a template of key question types as a starting framework

​
Advanced Techniques

​
Socratic Self-Reflection

Create prompts that guide AI through self-reflective questioning: 
---
provider: OpenAI
model: gpt-4o
temperature: 0.6
---

# Socratic Self-Reflection Process

## Initial Response:
First, I'll provide my initial thoughts on: {{ topic }}

My initial response is:

## Self-Questioning:
Now, I'll question my own thinking process:

1. **Clarity Check**: What concepts in my response need better definition?
2. **Evidence Examination**: What evidence supports my claims? What evidence contradicts them?
3. **Assumption Identification**: What assumptions am I making that I haven't justified?
4. **Perspective Analysis**: Whose perspectives might be missing from my analysis?
5. **Logical Consistency**: Are there any contradictions or tensions in my reasoning?
6. **Limitation Awareness**: What are the boundaries of my knowledge on this topic?

## Revised Response:
Based on this self-questioning process, I'll now provide a more nuanced response:

## Meta-Reflection:
Finally, I'll reflect on how this questioning process changed my thinking:

​
Nested Socratic Questioning

Implement layered questioning that progressively deepens inquiry: 
---
provider: OpenAI
model: gpt-4o
temperature: 0.7
---

<step>
# Primary Question Layer

## Central Question:
{{ central_question }}

## Initial Response:
Let me provide an initial response to this question...

## First-Layer Questions:
1. What are the key components of this question?
2. What background knowledge is needed to address it?
3. What approaches could be used to explore this question?
</step>

<step>
# Secondary Question Layer

For each first-layer question, let me ask deeper follow-up questions:

## On Key Components:
1.1. Why is component X essential to this question?
1.2. How do these components interact with each other?
1.3. Are there hidden components not immediately apparent?

## On Background Knowledge:
2.1. How reliable is our current knowledge in this area?
2.2. What knowledge gaps might affect our understanding?
2.3. Where might existing knowledge contain biases or blindspots?

## On Possible Approaches:
3.1. What are the strengths and limitations of each approach?
3.2. How might these approaches complement each other?
3.3. What novel approaches haven't been considered?
</step>

<step>
# Tertiary Question Layer

Let me select the most promising threads and explore even deeper:

## Deepening Thread 1:
[Select most significant question from previous layer]
- Why is this particular aspect critical?
- What would fundamentally change if we viewed this differently?
- How certain can we be about our understanding here?

## Deepening Thread 2:
[Select second most significant question]
- What underlying principles are at work here?
- How might this connect to other domains or concepts?
- What paradoxes or tensions emerge at this level?

## Synthesis Questions:
- How do these deepening inquiries relate to our central question?
- What new insights emerge from this layered questioning process?
- What fundamental reconceptualization might be needed?
</step>

<step>
# Integration and Response

After this nested questioning process, let me synthesize what we've discovered:

## Key Insights:
- [Key insight 1]
- [Key insight 2]
- [Key insight 3]

## Transformed Understanding:
How our understanding of the central question has evolved...

## Remaining Questions:
What important questions still remain for further exploration...

## Final Response:
My comprehensive response to the central question, informed by this layered inquiry process:
</step>

​
Integration with Other Techniques

Socratic questioning works well combined with other prompting techniques:
  • Chain-of-Thought + Socratic Questioning: Use questions to guide each step in a reasoning chain
  • Tree-of-Thoughts + Socratic Questioning: Question different branches of reasoning to evaluate alternatives
  • Self-Consistency + Socratic Questioning: Use consistent questioning patterns across multiple solution attempts
  • Role-Playing + Socratic Questioning: Have different expert personas ask questions from their perspectives
The key is to maintain the emphasis on thoughtful questioning that promotes discovery rather than simply directing toward answers. Explore these complementary prompting techniques to enhance your AI applications:

​
Critical Thinking Techniques

​
Structured Reasoning Approaches

​
Knowledge Enhancement

​
Collaborative Approaches