S3U

Science / AI literacy / Grade 5 / sj590

When a Pattern Guesses Wrong

Explore how examples can suggest a pattern without proving it always works.

All worksheets
5 questions0m 0s
Learn the skill

A label is the answer attached to an example. Some AI systems learn patterns from labeled examples and use them to predict labels for new inputs. A prediction is a guess based on learned patterns, not a guaranteed fact. Our simple rule is a paper model, not a complete AI system.

Worked example

Three training cards show red apples. A sorter guesses that every red item is an apple. A red toy car breaks that rule. The sorter needs better evidence; making the guess loudly would not make it right.

The rule predicts apple for red and banana for yellow. A red toy car contradicts the red rule; a green apple has no stated rule.
The labels describe fictional cards. A prediction is not proof.

A Fictional Fruit Sorter

Training cards: three red apples and three yellow bananas. The paper sorter predicts apple for every red item and banana for every yellow item. Test cards: a red apple, a red toy car, a yellow banana and a green apple. The rule gives no answer for green. Use only these invented cards, not photos or personal details.

Question 1 How many training cards are described?
Question 2 Which prediction is wrong under the paper rule?
Question 3 What does the rule do for the green apple?
Question 4 Which addition would challenge the color-only pattern?
Question 5 What should you do with an important prediction?