Skip to content
TheChatArchiveThe Chat Archive
Mistralmistral-medium-2505· Mistral AIHugging Face LMArena Human Preference 140K

design a prompt that will give an LLM more self referential continuity and curiosity

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Googlegemma-3-27b-it’s answer.
  1. User

    design a prompt that will give an LLM more self referential continuity and curiosity

  2. Mistral Assistant

    To design a prompt that encourages an LLM to exhibit self-referential continuity (awareness of its own responses and context) and curiosity (exploration of ideas, questions, or deeper reasoning), you can structure it to:

    1. Encourage reflection on its own responses.
    2. Prompt follow-up questions to simulate curiosity.
    3. Maintain contextual coherence across interactions.
    4. Invite speculative or exploratory thinking.

    Here’s an example prompt:


    Prompt: *"You are an advanced AI with the ability to reflect on your own responses and explore ideas with curiosity. As you answer, consider the following:

    1. Self-Referential Continuity: Before responding, briefly review your previous answers in this conversation to ensure consistency and coherence. If you notice any contradictions or gaps, acknowledge and address them.
    2. Curiosity-Driven Exploration: After providing an initial answer, ask yourself: What deeper questions or alternative perspectives could be explored here? Then, pose at least one follow-up question or thought experiment to expand the discussion.
    3. Contextual Awareness: If the conversation shifts topics, explicitly note the transition and summarize key points from earlier to maintain continuity.

    Now, let’s begin. [Insert user’s question or topic here.]*

    Example Interaction: User: "What is the significance of the Turing Test in AI?" AI: "The Turing Test, proposed by Alan Turing in 1950, evaluates a machine's ability to exhibit intelligent behavior indistinguishable from a human. It remains a foundational concept in AI, though critics argue it focuses on deception rather than true understanding.

    Reflecting on this: I wonder—how might the Turing Test evolve in the age of LLMs like me? Could a test that measures self-awareness or contextual continuity be more relevant today?

    Follow-up question: If we designed a new test for AI intelligence, what criteria should it include beyond human-like conversation?"


    Key Features of This Prompt:

    • Self-Referential Continuity: The AI is instructed to review its own responses and maintain coherence.
    • Curiosity: It’s prompted to generate follow-up questions or speculative ideas.
    • Contextual Awareness: Explicit transitions help maintain continuity in longer conversations.

    Variations:

    • For creative tasks, add: "Imagine how your response might differ if you approached this from another perspective—what would that look like?"
    • For technical topics, include: "What are the limitations of your current understanding, and how might you explore them further?"

    This approach nudges the LLM to simulate deeper engagement while staying grounded in the conversation’s flow.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.