K17

Knowing when not to use it

Make the decision deliberately, so that using it is a choice rather than a default.

When to use this: When something feels wrong
Learning impact: Strong change
Learning evidence: Credible with caveats
Where you are with the topic: Building understanding
How often to use it: Build and keep as a habit

When to use it

You are about to open the chat out of habit before trying anything alone.
The task is the assessed skill itself, not a side chore.
You want someone to argue the case for doing this without AI.

What to ask AI

Type something like: “I’m about to [task] for [purpose]. Before helping: is this a task where using you would undermine what I’m trying to build?”

Ask it to argue the case for you doing it alone, without being diplomatic.

Keep the hard part on your side: you decide, and sometimes you close the tab.

How it helps

The decision nobody examines is whether to use the tool at all.

It is open, fast, and asking costs nothing, so it becomes a default.

Some struggle is load-bearing: the ten minutes of being stuck are where structure gets built.

If what’s assessed is unaided performance, practising with help is practising the wrong task.

Sometimes the process is the product: writing to find out what you think, sitting with a problem.

Knowing when not to use it makes using it a choice again.

How to run it

Before opening the chat, name the task and the purpose in one line each.
Ask for an explicit case against using AI on this task.
If the case is “this is the assessed skill” or “struggle is the point,” close the tab.
If you proceed, set a constraint: attempt first, hints only, no full solution.
Afterward, note whether you could have done the core move alone.

Example exchange

You

I’m about to draft the main argument of my philosophy essay for a take-home exam that bans AI. Before you help: would using you undermine what I’m trying to build? Argue for me doing this alone, don’t be diplomatic.

Yes. The assessed skill is your unaided argument construction. If I generate the spine, you’re practising a different task. Write your outline alone for thirty minutes. If you come back, ask me only to pressure-test premises you already wrote.

You

Closing the tab. I’ll come back only for premise stress-testing.

That’s the method working: a deliberate no.

Copyable prompt

I'm about to [task] for [purpose]. Before helping: is this a task where using
you would undermine what I'm trying to build? Argue the case for me doing this
alone, and don't be diplomatic about it.

The Tell

Here is how you know this method has flipped: you can’t remember the last time you decided not to use it for something.

Then it isn’t a tool, it’s a default.

Interrupt that default once today with a deliberate no.

Principle evidence

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

The underlying learning idea is rated moderate. Appropriate help-seeking and judgments about cognitive offloading matter: some struggle is productive, and help that removes load-bearing difficulty can hurt learning. Principle moderate for when-not-to-ask; always-on AI makes the decision novel, so AI-specific evidence is still thin.

AI delivery evidence

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

Ironically, you may ask AI to argue against using AI. That prompt can help, but no delivery trial is claimed. The learning is your choice to stop.