PROMPT This AI Challenge

"The Hidden Battle Over Books and AI — Why It Matters to Every Creator" featuring Julie Trelstad

Every episode of the PROMPT This podcast includes an AI Challenge for the audience.  Follow the instructions below to complete this episode's challenge.

AI answers often sound confident. That does not mean they are correct.

You do not need a complex system to use AI safely and avoid bad answers, or hallucinations. You need a few simple habits that help you understand where an answer came from and whether you should trust it.

Start With a Topic You Know Well

Pick something you already understand deeply. This matters because you need a mental “truth meter” to compare against the AI’s answer.

Examples:

• What lead qualification questions to ask a prospect of your company

• How your company's manufacturing metric is actually calculated

• How your company's service is delivered end to end

• How pricing or packaging decisions for your products get made
  
  

Prompt 1: 

Explain [this topic] to me like I am new to the field.

Read the answer and notice:

• What is right but basic

• What is missing

• What feels off

That reaction is your baseline.

The 5-Step AI Source Detective Process

Step 1: Ask What the Answer Is Based On

Goal: Separate facts from guesses.

Prompt: 

For each part of your answer, tell me whether it is:

• Common knowledge

• Based on my prompt

• An assumption

• A guess or uncertain

If the AI cannot clearly label its own answer, that is a warning sign.

Step 2: List the Main Claims

Goal: Turn paragraphs into checkable pieces.

Prompt: 

List the main factual claims in your answer as short, one-sentence statements.

This makes weak spots obvious and prevents vague language from hiding problems.

Step 3: Force the AI to Admit What It Does Not Know

Goal: Reduce made-up details.

Prompt: 

Rewrite the answer. If you are unsure about something, say so clearly. Separate what is generally true from what depends on context.

A safer answer is better than a confident one.

Step 4: Make the AI Criticize Itself

Goal: Pressure-test the answer.

Prompt: 

Point out the biggest mistakes, oversimplifications, or missing warnings in your answer.

If the answer falls apart under criticism, do not use it as-is.

Step 5: Decide What You Would Verify

Goal: Turn doubt into action.

Prompt: 

For the top claims, explain what I would need to check to confirm them and where I would look.

Now you know what matters and what needs proof.

The One-Minute Checklist

Before using an AI answer in real work, ask yourself:

• Do I know what this answer is based on?

• Can I list the main claims?

• Did the AI admit uncertainty?

• Did I challenge the answer?

• Do I know what I would verify?

If you do just two things, do these:

1. Ask where the answer came from

2. Ask the AI to criticize itself

Final Thought

AI is best treated like a fast assistant, not an expert witness. If you slow it down just enough to ask where its answers come from, challenge them, and decide what needs proof, you can use AI with confidence instead of hope. That is what it means to be an AI source detective.