The Netyeti's Journal

Asking Stupid Questions: A Guide to AI Prompting for Students

5 min read

Asking Stupid Questions: A Guide to AI Prompting

The NetYeti Philosophy

“If you are too shy or afraid to ask that question which is deemed too obvious (aka common sense), you are leaving a hole in your knowledge. AI is the perfect place to ask the stupid questions so you can get the smart answers. Don’t let a ‘manageable mess’ turn into a catastrophe just because you were afraid to look ‘new’.”

The Instructor’s Perspective

Prompting isn’t about being a “wizard.” It’s about context and intent. It’s the difference between asking “How do I fix my computer?” and “I have a Linux laptop that won’t boot past the GRUB screen. Here is what I see…”

The Framework: C-R-E-A-M

(Context, Role, Example, Action, Mindset)

1. Context (The “Why”)

Don’t just ask a question. Tell the AI what you are doing.

  • Bad: “How do I use ls?”
  • Good: “I’m a new Linux student at BTC trying to understand file permissions. I’m using the ls -l command in my home directory and I’m confused about the drwxr-xr-x string.”

2. Role (The “Who”)

Give the AI a persona. It changes the “voice” of the answer.

  • Bad: “Tell me about networking.”
  • Good: “Act as a Senior Network Instructor (and Army Veteran). Explain the OSI model using a PACE plan for communication reliability.”

3. Example (The “What”)

Give the AI a pattern to follow.

  • Good: “I want you to summarize this man page. Here is an example of the format I like: [Command Name, Primary Flag, Example Usage].“

4. Action (The “How”)

Be specific about what you want.

  • Bad: “Fix this code.”
  • Good: “Review this Bash script for common pitfalls, especially around variable quoting and error handling. Suggest a more ‘NetYeti’ way to handle the logic.”

5. Mindset (The “NetYeti” Twist)

Tell the AI to challenge you.

  • Good: “After you explain this, ask me a Knowledge Check question to see if I understood it. Don’t be afraid to tell me if my approach is a ‘stupid question’ in the wrong direction!”

The Local-First Advantage (Pillar of Learning)

Why do we prompt locally on Ollama before hitting the cloud?

  1. Privacy: You can ask the really stupid questions without them being logged in a corporate data center.
  2. Speed: Instant iteration. If the prompt fails, tweak it and go again. No tokens, no cost.
  3. Exploration: You are the “Admin” of the brain. You can see the model load, watch the VRAM usage, and understand the mechanics of the intelligence.

NetYeti Tactical Prompting: The Multi-Turn Strategy

Don’t try to get the “perfect” answer in one go. That’s a Common Pitfall. Use a multi-turn approach to build a “Knowledge Bridge.”

Turn 1: The Briefing (Set the Stage)

“I am a student at BTC working on a Linux Home Lab. Act as a Senior Technology Instructor (The NetYeti). I want to learn about [Topic, e.g., SSH Keys]. Before we start, tell me the Primary (P) concept I need to understand.”

Turn 2: The Deep Dive (Ask the “Stupid” Question)

“Okay, I understand [Topic]. Now, here is the ‘stupid question’: Why do I need a passphrase if the key is already ‘secure’? Explain this using a Teachable Moment about a ‘Manageable Mess’.”

Turn 3: The Practical (The AAR)

“Write a step-by-step guide for me to generate an Ed25519 key. After each step, explain WHY we are doing it. At the end, give me a Knowledge Check.”

The Troubleshooting PACE Plan

When you are stuck on a technical problem, use this prompt structure:

  • Primary (P): “Here is exactly what I am trying to do (The Mission).”
  • Alternate (A): “Here is what I have already tried (The AAR).”
  • Contingency (C): “Here are the errors I’m seeing (The Intelligence).”
  • Emergency (E): “Tell me what the most common ‘stupid mistake’ is for this scenario.”

The “NetYeti” Teachable Moment

Manageable Mess vs. Catastrophic Failure

A Manageable Mess is when you try something, it breaks, but you know why it broke (or have the logs to find out). AI is great at helping you turn a mess into a lesson. Catastrophic Failure is when you copy-paste code you don’t understand and it deletes your /home directory.

Always ask the AI: “Explain this code to me like I’m 5 before I run it.”

Common Pitfalls

  • Blind Trust: Never copy-paste AI code without reading it. If you don’t understand it, it’s not a “win,” it’s a “liability.”
  • The Vague Voids: Vague prompts get vague answers. If the AI is giving you junk, look at your context first.
  • Signal Loss: If the AI is “hallucinating” (making stuff up), it’s usually because you didn’t give it enough guardrails.

Knowledge Check

  • Why is it better to tell the AI “who it is” (Role) before asking a question?
  • How does providing an “Example” improve the output of an LLM?
  • What is the difference between a “Manageable Mess” and a “Catastrophic Failure”?

Related: AI Command Center, The AI Homelab Story, AI Education Project

Graph View

Backlinks