# Habit 4: Think Win-Win → Co-Create with the Model #### What It Means in LLM Context Covey’s original Habit 4 is about **mutual benefit**, not competition or extraction. In LLM interaction, this means: Don’t treat the model as a magic answer box.\ Instead, approach it as a **thought partner**. One that brings patterns, options, and scaffolding, while **you bring intent, values, and judgment**. The best sessions feel like **dialogues**, not transactions. *** #### Common One-Sided Patterns | User Posture | Model Effect | Risk | | ---------------------------------- | ------------------------------------ | ---------------------- | | Treats LLM as a search box | Shallow autocomplete responses | Surface-level insights | | Uses prompt tricks with no context | Theatrical outputs, little substance | Output feels fake | | Talks at the model, not with it | No adaptive *feedback loop* | Stagnant results | | Dumps data without role setting | Overwhelming or misaligned synthesis | Misweighted priorities | *** #### Co-Creation Mindset A collaborative prompting session follows a rhythm like this: **1. You declare intent** “I’m starting my senior project on drone applications. I’m torn between building a prototype, writing a simulation of autonomous search behavior, or designing a community-focused deployment plan. I want help clarifying the most compelling direction that balances feasibility and impact.” **2. Model offers scaffolding or provocation** “Great — let’s break this down. You could frame your project through three different lenses: * Engineering prototype – a functioning drone with basic sensor response * Algorithmic focus – simulate multi-agent pathfinding in virtual disaster zones * Humanitarian design – create a deployment model for underserved communities Would it help to compare these by effort, originality, and stakeholder impact?” **3. You clarify, correct, or build on it** “Yes, but I also want to include some AI work, not just design. I don’t have access to hardware this semester, so prototyping might be out. Can we explore what kind of simulations I could build that would be respected by faculty, something that shows real thinking, not just a flashy interface?” **4. Model adapts, deepens, or reframes** “Understood. Given no hardware access, here are three simulation-based paths you could consider: * **Swarm Search Simulation:** Build an environment where multiple agents (drones) search for survivors, using AI-driven heuristics like A\* or potential fields. * **Obstacle Navigation AI:** Simulate a single drone navigating through unstable structures, optimizing for time and energy use. * **Deployment Strategy Visualizer:** Create a simulation that shows how varying community infrastructures affect rescue efficiency (e.g., rural vs. urban). I can help scaffold each one — which direction feels closest to your vision?” **5. You summarize insight and redirect next turn** “I love the swarm idea, especially if I can compare different search algorithms under time pressure. Let’s go deeper into how I might model that environment. What tools should I use, and how do I keep it simple but credible?” This isn’t “prompt → answer”. It’s **move → countermove → synthesis.** *** #### Co-Creation Prompts in Action | Intent | Prompt | | --------------- | ---------------------------------------------------------------------------------------- | | Brainstorming | “Let’s explore 3 radically different angles before narrowing in.” | | Thought partner | “Here’s what I think so far. What might I be missing?” | | Live refining | “Rewrite this paragraph while keeping the tone, but don’t remove the emotional tension.” | | Stress-testing | “Take this answer and flip it. What would a skeptic say?” | | Diagnostic mode | “Help me identify whether my logic breaks down anywhere.” | *** #### Real Examples: Win-Win vs Win-Lose **Win-Lose Prompt (Extraction):** “Summarize this research and give me a metaphor.” **Win-Win Prompt (Co-Creation):** “Here’s a rough metaphor I was playing with after reading this research. Can you test it for accuracy and suggest ways it could become more intuitive to a high school student?” *** **Win-Lose Prompt (Overload):** “Here’s 3 pages of notes. Organize and summarize.” **Win-Win Prompt (Framed):** “These notes were from a conversation with an elder about cultural loss. I want to preserve the nuance and emotion while making it digestible to a 20-year-old student of indigenous studies. Help me find the emotional throughline before organizing.” *** #### Self-Check * Did I treat this exchange as a dialogue or a delivery? * Am I allowing the model to surprise me, or am I overcontrolling? * Did I offer my own signal for the model to work with, or just expect brilliance from a blank slate? *** #### Win-Win is Not Flattery — It’s Friction + Trust You bring the **values, judgment, and direction.**\ The model brings **breadth, reactivity, and scaffolding.** Neither is complete alone. The true “win-win” emerges when you embrace the model’s weirdness as **provocation**, not failure. *** #### Integration Cue *Before each prompt, ask: “What is my role in this co-creation, and what is the model’s?”* The more you design the relationship **dynamically**, the more potent the results. *** #### Metaphor Don’t treat the model like a chef preparing your meal.\ Treat it like a **kitchen**, full of tools, ingredients, and ideas. You bring the taste, the memory, and the occasion.\ Together, you make something worth serving. *** --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://praxistutor.gitbook.io/llm/reframing-the-7-habits-for-reflective-llm-use/section-2-the-seven-habits-of-reflective-llm-use/habit-4-think-win-win-co-create-with-the-model.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.