Completion problems

Receive a solution with the last steps missing, and finish it.

When to use this: While you practise
Learning impact: Strong change
Learning evidence: Strong research tradition
Where you are with the topic: Building understanding
How often to use it: Worth repeating often

When to use it

You have twenty minutes tonight and need real practice, not more highlighting.
Homework is done, but you still want one honest attempt before bed.

What to ask AI

Type something like: “Give me the same type of problem four times:”.

Ask it to generate material for you to attempt, problems, cues, or a schedule, not finished answers.

Keep the hard part on your side: you retrieve, attempt, explain, or decide before you ask for help.

How it helps

Worked examples are efficient for novices and stop helping as you improve (expertise reversal).

Unaided problems are where you need to end up and are too hard immediately after a first exposure.

The gap between them is where students stall, they can follow every solution and start none.

Completion problems are that gap, made into a sequence.

They preserve the worked-example benefit for the part you're still shaky on while giving you real generative practice on the part you're ready for.

And because the missing portion grows, the support fades, which is the half of scaffolding almost everyone omits.

A single completion problem is a mild exercise.

A sequence of four with increasing gaps is a structured route from "I can follow it" to "I can do it" in one sitting.

How to run it

Name a concrete target for completion problems (narrower than a whole chapter).
Use the copyable prompt below, filling in your topic.
Do the cognitive work yourself before you ask AI to continue.
Stop when you can point to what you can now do that you could not do before.

Example exchange

You

Give me a completing-the-square problem with the last two steps blank. Don't fill them, wait for me.

Here's the start: x squared plus 6x plus 2 equals 0 becomes (x plus 3) squared minus 9 plus 2 equals 0. Finish from here.

You

So (x plus 3) squared equals 7, then x plus 3 equals plus or minus root 7, so x equals negative 3 plus or minus root 7.

That completes it cleanly. Ready for one with only the first step given?

Copyable prompt

Give me the same type of problem four times:
1. fully worked, with the decision points labelled;
2. worked except the last two steps;
3. only the first step given;
4. nothing.
Don't move on until I've completed each one. Tell me where I stall.

The Tell

Here is how you know this method has flipped: you complete every version and can't do a fresh problem the next day. The scaffold was doing more work than you noticed, go back and ask for version 4 first.

In that moment the AI (or the schedule, or the story) did the thinking, and you only recognised a finished product.

Tighten the prompt for completion problems so you attempt, decide, or retrieve before anything is handed to you.

Principle evidence

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

The underlying learning idea is rated strong. Completion / faded worked examples sit inside CLT and example-based learning (van Merriënboer, Renkl, Sweller tradition): learner finishes missing steps, reducing extraneous load while preserving generative effort. Principle strong for novices; AI-produced partial solutions need correctness checks, delivery speculative.

AI delivery evidence

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

The learning principle may be strong, but that does not prove an AI session delivers it well. Treat AI delivery as speculative unless a study of tutoring with this method is named, and none is claimed here.