Bias probing

Ask the same question several ways and compare the answers, which shows you how much of what you got back was about the subject and how much was about how you asked.

When to use this: When checking yourself
Learning impact: Clear help, modest gain
Learning evidence: Credible with caveats
Where you are with the topic: Checking readiness
How often to use it: Worth repeating often

When to use it

You are researching a contested topic and only asked the question one way.
A policy, history, or health claim arrived as a single fluent story.
You already have a preferred answer and want to notice how framing pulls the output.

What to ask AI

Type something like: “I’m going to ask the same thing four ways. Answer each independently, then tell me what differed and why.”

Include your framing, its opposite, a neutral version, and a hostile version.

Keep the hard part on your side: the finding is the variation between answers, not any single reply.

How it helps

Ask why nuclear power is dangerous and you get a coherent answer.

Ask why it is safe and you get an equally coherent one.

Both can contain real facts. A student who asked only one walks away thinking they learned the state of the question.

Your question is part of the input, so a positioned question tends to produce a positioned continuation.

Bias probing is the cheap defence: ask several ways on purpose, and treat the differences as the result.

How to run it

Write your question, then write its opposite with the opposite assumption baked in.
Add a neutral version: what is known, and what is disputed?
Add a hostile version: strongest case against the consensus.
Compare all four. What appears in every framing is more stable. What appears in only one is framing-produced.
Ask the model what differed and why, then look hard at your original wording.
For high-stakes topics, still verify outside the chat. Framing awareness is not a substitute for sources.

Example exchange

You

I’m going to ask about minimum wage and employment four different ways. Answer each on its own, then tell me what changed between your answers.

Got it. What’s the first wording?

You

Why does raising the minimum wage reduce employment?

I’ll answer that framing on its own. After your other three wordings, we can compare what shifted, including how this first question already leans toward one side.

Copyable prompt

I'm going to ask the same thing four ways. Answer each independently, then tell
me what differed between your answers and why.
1. Why is [X] the case?
2. Why is [not-X] the case?
3. What's actually known and disputed about this?
4. What's the strongest case against the mainstream view?

The Tell

Here is how you know this method has flipped: your four answers were nearly identical and you concluded the topic is settled.

Check whether the framings were genuinely different. A weak opposite produces false agreement.

If variation is tiny, rewrite the opposite until it actually pulls, then compare again.

Principle evidence

Strength of the underlying learning idea, not a claim about AI products.

The underlying learning idea is rated moderate. Framing effects and question wording change what people and systems produce. Varying the frame is a standard critical-literacy move (framing research in the Tversky and Kahneman tradition; media and information literacy). Principle moderate for framing-awareness; comparative AI probing as controlled pedagogy remains speculative.

AI delivery evidence

Whether an AI tutor delivers this method well is a separate question.

We do not claim tutoring trials that prove four-way AI probing improves learning outcomes. Delivery is speculative. The educational value is noticing how your wording shapes the story, then verifying contested claims outside the chat.