Skip to main content
Updated Mar 15, 2026

The Prediction Lock

AI quality is downstream of question quality. The student who asks better questions gets better answers from every tool, every person, and every system for the rest of their career.

Most students type the first thing that comes to mind into an AI tool and accept whatever comes back. This chapter trains you to treat question formulation as a disciplined practice. A vague question produces a vague answer. A precise, layered question produces insight. This is not about prompt engineering — it is about thinking clearly enough to know what you actually need to know.

Exercise 1: The Prediction Lock​

Layers Used: Layer 1 (Predict Before You Prompt), Layer 2 (Reasoning Receipt)

What You Do​

You receive a scenario describing an unexpected outcome. Before touching any AI tool, write down in a sealed document:

  1. (a) Your initial diagnosis of what went wrong
  2. (b) The 10 most important questions you would ask to understand this problem, ranked by diagnostic power
  3. (c) Your predicted answer to each question

Timestamp and submit this prediction lock. Only then open claude.ai or chatgpt.com. Feed your top 5 questions to both tools. Document every response, noting whether you accept, reject, or modify each AI answer.

Choose Your Scenario​

Scenario A (Business): "A retail company's online sales dropped 15% despite a 20% increase in marketing spend."

Choose one. The exercises work identically regardless of which you pick.

Your Deliverable

A sealed prediction document (timestamped before AI use) containing your diagnosis, 10 ranked questions with predicted answers, followed by a reasoning receipt showing all 5 prompts sent to both Claude and ChatGPT, the responses received, and your accept/reject/modify decision for each with a one-sentence justification.

AI Check Prompt -- Copy and paste into claude.ai or chatgpt.com
I am a student learning question formulation. Below is a business scenario,
followed by my initial diagnosis and 10 ranked diagnostic questions.
Please evaluate:

(1) Rate each of my 10 questions on a scale of 1-10 for diagnostic power --
    how likely is this question to reveal the root cause?
(2) Identify which of my questions are too vague, too narrow, or redundant.
(3) Suggest 3 questions I missed that would have been more diagnostic than
    my weakest 3.
(4) Evaluate my ranking -- did I put the highest-value questions at the top?
(5) Rate my overall question formulation skill from
    Beginner / Developing / Proficient / Advanced and explain why.

Here is the scenario: [paste scenario].
Here is my work: [paste your prediction lock document].

Finally, complete the Thinking Score Card for this exercise:
Independent Thinking (1-10), Critical Evaluation (1-10),
Reasoning Depth (1-10), Originality (1-10), Self-Awareness (1-10).
For each score, give a one-sentence justification.
Deliverable Template (click to expand)

PREDICTION LOCK TEMPLATE

  • Date/Time: ___
  • Scenario: [paste]
  • Section A - My Diagnosis (2-3 sentences): ___
  • Section B - My 10 Questions (ranked):
    • Q1 [highest value]: ___ | Predicted answer: ___
    • Q2: ___ | Predicted answer: ___
    • Q3: ___ | Predicted answer: ___
    • Q4: ___ | Predicted answer: ___
    • Q5: ___ | Predicted answer: ___
    • Q6: ___ | Predicted answer: ___
    • Q7: ___ | Predicted answer: ___
    • Q8: ___ | Predicted answer: ___
    • Q9: ___ | Predicted answer: ___
    • Q10: ___ | Predicted answer: ___
  • Section C - REASONING RECEIPT:
Prompt #Prompt SentToolResponse SummaryDecisionJustification
1ClaudeAccept/Reject/Modify
2ChatGPTAccept/Reject/Modify
3ClaudeAccept/Reject/Modify
4ChatGPTAccept/Reject/Modify
5ClaudeAccept/Reject/Modify
6ChatGPTAccept/Reject/Modify
7ClaudeAccept/Reject/Modify
8ChatGPTAccept/Reject/Modify
9ClaudeAccept/Reject/Modify
10ChatGPTAccept/Reject/Modify

What This Teaches You​

You learn that the quality of your questions determines the quality of every answer you will ever get — from AI, from colleagues, from data. By predicting first and then comparing, you see exactly where your thinking was strong and where it was lazy. The AI grading reveals blind spots in your questioning that you cannot see yourself. Over time, you internalize what makes a question diagnostic rather than decorative.

Last updated on