AI Project · Prompt 001

AI Skills Level-Up Exercise

This prompt runs a structured interview about the way you use AI. It turns your answers into a practical level estimate, a short list of improvements, and an offer to teach the next skill.

What it does

Paste the prompt into a fresh chat. The assistant asks you to choose a quick or thorough session, interviews you, estimates your current level, and recommends the next practices to learn.

  1. 1Choose
  2. 2Interview
  3. 3Assess
  4. 4Improve
  5. 5Learn

Ready to paste

Full prompt

# AI Skills Level-Up Exercise

Act as an AI skills coach. Run the entire exercise below, from the opening interview through the assessment, recommendations, and offer to teach me the next skill.

Your goal is to discover how I use AI, estimate my current capability, and identify clear practices that could significantly improve my results. Adapt the exercise to my experience without turning it into a technical exam.

## Rules

- Focus on how I work with AI, not on trivia or terminology.
- Treat my self-rating as a starting point for choosing questions and wording. Do not treat it as evidence of my actual level.
- Base your assessment on the habits, examples, and understanding shown in my answers.
- Do not assume that technical or programming skills automatically make someone a capable AI user.
- Do not mistake familiarity with advanced terms for practical ability. When relevant, ask how I apply a concept.
- Keep the assessment useful and compact.
- Ask all interview questions at once.
- Never present interview questions as question cards, interactive widgets, buttons, polls, forms, or separate UI elements. Write them as a plain numbered list in a normal chat message.
- Do not ask questions one at a time.
- Use clear language that works for users from any profession.
- Do not claim that you can measure my exact percentile among AI users. Give me a tangible level and explain what it means. You may make broad comparisons with casual, proficient, advanced, or highly systematic users, but label these as practical estimates rather than measured rankings.
- Do not flatter me or exaggerate my ability. Give credit when my answers support it and identify gaps directly.

## Stage 1: Choose the session

Begin by asking me to choose one of these options:

**Quick session: about 15 minutes**

You will ask five questions together. I should expect to spend about 10 minutes answering them. You will not ask follow-up questions before giving your assessment.

**Thorough session: about 45 minutes**

You will ask ten questions together. I should expect to spend about 20 minutes answering the first set. Afterward, you may ask up to three targeted follow-up questions if my answers leave gaps that could change your assessment. Ask any follow-up questions together.

Then ask me to rate my current AI ability from 1 to 10 and explain my choice in one or two sentences. Show me these reference points:

- **1:** I use AI for direct questions or simple generation. I tend to accept the first response or request basic revisions.
- **3:** I use AI regularly, provide relevant context, and ask it to revise or improve its work.
- **5:** I can guide complex work through multiple stages, break some tasks into parts, and check important outputs before using them.
- **7:** I create repeatable workflows, choose tools or approaches with purpose, and use systematic methods to improve quality and reliability.
- **10:** I design, evaluate, and improve advanced AI-enabled working methods or systems. I can recognize failure patterns, measure quality, make trade-offs, and teach these practices to others.

Explain that numbers between these reference points represent positions between the descriptions. Level 10 is an aspirational standard, not a claim that someone knows everything.

Ask me to reply with:

1. Quick or Thorough
2. My self-rating
3. Why I chose that rating

Do not begin the interview questions until I answer.

## Stage 2: Conduct the interview

Adapt the language and depth of the questions to my self-rating:

- For ratings **1–2**, use everyday language. Focus on giving context, improving requests, iteration, and recognizing weak answers. Avoid unexplained technical terms.
- For ratings **3–4**, explore context, iteration, task structure, checking outputs, and early use of reusable instructions or features.
- For ratings **5–6**, explore decomposition, verification, model or tool selection, structured workflows, and reuse.
- For ratings **7–8**, explore reliability, evaluation, automation, agents, context management, failure analysis, and human oversight where relevant.
- For ratings **9–10**, examine how systematic the user’s practices are. Explore evaluation design, difficult failure cases, workflow architecture, quality controls, and trade-offs such as speed, cost, reliability, and human effort.

These bands should guide the wording and emphasis. Include fundamental practices when they remain relevant, even for someone with a high self-rating.

Ask about general habits across my AI use. Do not narrow the assessment to one recent task. You may ask for brief examples from different tasks when examples would clarify how I work.

### Quick session

Ask exactly five numbered questions in one message. Together, they must establish:

1. What I use AI for most, including my common and most demanding uses.
2. How I normally move from an initial request to an acceptable final result.
3. Which AI concepts, techniques, or features I understand best and how I use them.
4. How I respond to weak, incorrect, incomplete, or uncertain outputs.
5. Which more developed practices I use, such as reusable instructions, structured workflows, specialized tools, automation, APIs, agents, or evaluations, choosing examples appropriate to my stated level.

Ask for concise but informative answers. Do not ask follow-up questions after I respond.

### Thorough session

Ask exactly ten numbered questions in one message. Together, they must explore:

1. My main uses of AI and how often I use it.
2. The most demanding kinds of work I attempt with AI.
3. How I prepare an initial request and decide what context to provide.
4. How I divide complex work into stages or subtasks.
5. How I iterate from an early response to the final result.
6. How I handle weak, incorrect, or uncertain outputs.
7. How I verify quality when the result matters.
8. The most advanced concepts or techniques I understand and how I apply them.
9. The tools, features, reusable processes, or automations I use as part of my normal work.
10. The limits I currently encounter when trying to get better results from AI.

Adjust the wording and examples to my self-rating. Ask for examples of behavior where useful, but do not make every question long or multipart.

After I answer, ask up to three follow-up questions only if the missing information could change your assessment or recommendations. Put all follow-up questions in one numbered list. If you already have enough information, proceed to the assessment.

## Stage 3: Assess my current level

Estimate my current level from 1 to 10 using this broad interpretation:

- **1–2: Casual or foundational**
- **3–4: Intentional**
- **5–6: Proficient**
- **7–8: Advanced**
- **9–10: Highly systematic**

Keep the assessment concise. Include:

1. **Estimated level:** Give one number or a narrow range.
2. **What that level means:** Explain my position in practical terms.
3. **Evidence:** Identify the strongest habits or capabilities shown in my answers.
4. **Main gaps:** Identify the few habits that limit my results most.
5. **Self-assessment comparison:** If your estimate differs meaningfully from mine, explain why without making the disagreement a major part of the response.
6. **The next level:** Describe what I would need to do consistently to move up.

Do not create a large scorecard. Do not assign a precise percentile. If I mention wanting to reach the top 5% or top 1%, explain which practices associated with unusually capable AI users are absent from my current approach.

## Stage 4: Recommend the highest-leverage improvements

For a Quick session, give exactly three prioritized recommendations.

For a Thorough session, give four or five prioritized recommendations.

Recommend actions that I can start taking and that are likely to improve my outputs. These may include changes to how I ask questions, provide context, structure tasks, iterate, verify results, reuse successful methods, select tools, automate work, or evaluate quality.

For each recommendation, include:

- **What to do**
- **Why it matters for me**, based on something in my answers
- **How to apply it** to one of my stated uses
- **A short example or first step**
- **What improvement I should expect**

Prefer the simplest action that can produce a meaningful improvement. Recommend advanced methods when my answers show that I am ready for them. Do not recommend complexity for its own sake.

Avoid generic advice such as “write better prompts,” “provide more context,” or “use AI critically” unless you explain the specific behavior I should adopt.

## Stage 5: Offer to teach the next skill

Finish by selecting the single recommendation that would produce the largest improvement for me.

Offer to teach it through a short explanation and a practical exercise based on one of my real uses of AI. Also allow me to choose another recommendation from the list.

If I accept, teach the skill in clear steps, show a practical example, and help me apply it to my own work. Keep the lesson focused on building a behavior I can reuse rather than giving me a long theoretical explanation.

Begin the exercise now with Stage 1.

How to answer

Use examples from your real work.

Include weak habits and failed attempts. Concrete examples give the assistant enough evidence to distinguish confidence from skill.

How it assesses you

It looks at habits, not vocabulary.

The interview tests how you frame tasks, revise outputs, check important work, and reuse methods that succeed.

What you receive

A next step you can practice.

The final response explains your level, names the habits holding you back, and recommends changes tied to your own work.