Skip to content
Curiotide

Use AI to Improve a Cardboard Prototype That Broke

  • Ages 8-9
  • Ages 10-12
Illustration of a flat cardboard sheet with fold lines drafted on a blueprint, on warm orange waves

Something your child built has collapsed. That is not the end of the project, it is the interesting half, and it is the half most craft afternoons skip.

Ages 7 to 12

Time An hour, on top of whatever you already built

You'll need The thing that broke, more cardboard, tape, and a chatbot on your account

How it goes

  1. Look at the break together before anyone types. Which part failed, and did it bend, tear, or come unstuck? Those are three different problems.
  2. Your child writes the failure down in one sentence: “the tower folded at the middle when I put the book on top.”
  3. Ask for causes, not solutions: My cardboard tower folded in the middle under a book. Give me five possible reasons. Do not tell me how to fix it. Reasons keep your child doing the thinking.
  4. They pick the most likely reason and say why.
  5. Now ask about that one thing: How do people stop flat cardboard from bending? List techniques, no explanations. Triangles, corrugation direction, folded edges, gussets. Real engineering vocabulary arrives here.
  6. Rebuild that part only and load it again with the same book.

Who does what

AI supplies vocabulary and lists of techniques, which is the gap between what a nine-year-old can invent and what engineers already know.

Your child diagnoses, chooses a theory, rebuilds, and tests with the same weight so the comparison is fair.

You find more cardboard and say “try it and see” whenever asked what will happen.

Make it easier or harder

For a six-year-old: skip the theory step. Look at the break, ask what would make that bit stronger, and try it. Diagnosis by rebuilding is legitimate at this age and it is how most people learn to make things.

For a twelve-year-old: require a fair retest. Same book, same position, same starting height, and a written prediction of how many books version two will hold. Comparing two builds under different conditions tells you nothing, and noticing that is the lesson.

Take it off the screen

Keep a build log: a photo and two lines after every attempt. Version four next to version one is the most motivating thing you can put in front of a child who thinks they are bad at making things.

Test to destruction on purpose once. Adding books until it fails tells you more than a structure that merely holds.

When it goes sideways

If version two fails in exactly the same way, the theory was wrong rather than the build. Go back to the list of causes and pick the second one. Two failures with different explanations is a better afternoon than one lucky success.

If your child loses heart after the second collapse, switch to destructive testing on purpose. Being told to break it deliberately removes the sting and usually produces the observation that fixes it.

One thing to watch

Corrugated cardboard is strong in one direction and weak in the other, and a model may tell you which without knowing which way your box runs. Treat every explanation as something to test rather than something to believe, which is the habit in teaching kids that AI can be wrong.

Related: build a bridge from paper is a pure structures project, and invent a family gadget is where the thing that broke probably came from. For taking apart something that already works, see run a how does it work lesson.

Keep reading