AI Pair Programming Etiquette

1. Concept Overview & Systemic Problem

In traditional development, there is a practice known as Pair Programming: two live engineers sit at one computer. One ("Driver") types code on the keyboard, while the other ("Navigator") observes the screen, suggests ideas, looks for errors, and considers architecture.

In the era of large language models, this method has been revitalized. Now, your constant partner is artificial intelligence.

However, the quality of this collaboration directly depends on your AI Pairing Etiquette:

• If you treat the model like a dumb servant, throwing dry phrases like "make it do", ignoring errors, and getting frustrated — the results will be terrible.
• If you interact as two polite professionals — the speed and satisfaction of working together increase exponentially.

The key engineering principle: the art of being a good navigator for your digital pilot.

2. Poor Communication vs. Professional Partnership

TOXIC COMMUNICATION (Dead End):
User: "YOU BROKE THE SITE, NOTHING WORKS, FIX IT IMMEDIATELY!!!"
    └── The model sees panic without details, chaotically alters random code,
        and breaks the project even further.

─────────────────────────────────────────────────────────────

PROFESSIONAL PAIRING ETIQUETTE (Quick Resolution):
User: "Thank you, the layout looks nice. However, clicking the button triggers a TypeError: user is undefined on line 24.
Here’s the content of the props object: {...}. What do you think is causing user to come up empty?"
    └── The model instantly identifies the issue: "User is not loading from the API in time.
        Let's add a loading state check. Here’s the exact diff:"

3. Four Rules of Pair Programming Etiquette

1. Provide Precise Feedback (Log-Driven Feedback): don't write "not working" — copy the exact error message from the terminal or browser console.
2. Request Explanations, Not Just Code: add to your request "Explain in 2 sentences why you chose this approach." This enhances your own understanding of the system.
3. Praise Successful Solutions: a brief "Great, that worked, let's move on" reinforces a successful branch of context in the dialogue.
4. Timely Update the Chat: if the discussion has stretched over 50 messages — thank the model, copy the final summary, and start a fresh clean dialogue.

4. Production Engineering Scenarios

01. Effective Feedback Loop

Utilize precise error messages and context to guide the model towards accurate solutions, enhancing the collaboration quality.

02. Navigating Complex Problems

When facing intricate issues, employ the Reverse Question principle to clarify requirements and avoid miscommunication.

03. Maintaining Contextual Clarity

Regularly update the chat to prevent context overload, ensuring that the model operates with the most relevant information.

5. Pitfalls, Common Mistakes & Security

AI learns from human behavior. A calm, structured, respectful, and precise communication style makes the model your most dedicated and productive ally in any project.