AI Mistakes and Limitations
Machines make mistakes; they only know what they've been shown; bad training data leads to bad results; AI is not magic — just maths on data; showing edge cases and failures
Lesson
Set out your lesson
“AI Mistakes and Limitations means practising the idea at the table with a simple screen-free or sandbox check. We will demo once, then your child tries.”
Bring to the table
- Two short recommendation cards or screenshot printouts
- Pencil for compare notes
Extra for the adult
Worked example. A feed keeps showing football because past clicks were football — that can hide other topics (filter bubble).
Optional challenge. For AI Mistakes and Limitations, add one harder example after success.
What to notice. Keep one parent note: examples tried; results; support used. This session samples AI Mistakes and Limitations, not mastery.
Share and support
Family records and proof options
Keep a record of your learning
Save photos, notes, and attendance. Export a proof packet for compliance or portfolio reviews.
Practice this concept
Start practiceAI Mistakes and Limitations practice
Practice key ideas from AI Mistakes and Limitations.
Free · no login · no ads · Practice answers and completion are not saved. Parent observation remains the lesson record.
Learning resources
Extra practice links — they do not replace Gakuva’s teaching approach.
Code.org · Practice link — not the full lesson
AI for Oceans
Before you open · source notes
Deliberately probe mistakes: train a limited model, then test odd examples to see confident wrong answers. Discuss that AI can fail without “understanding.”
Free · no login required to play · account optional to save · no ads observed
Before you openSave a screenshot of a wrong or uncertain classification and a parent note of why the learner thinks the model slipped.
Opening this link does not save learning or assess mastery.
Source, access and review notes
Unknown — link only
Google Creative Lab · Practice link — not the full lesson
Teachable Machine · Image
Before you open · source notes
Adult-guided mistake lab: train with unbalanced or similar-looking classes, then show a misclassification and improve the examples.
Free · no login · no ads observed · webcam optional
Before you openSave a misclassification screenshot and a second screenshot after adding better training examples.
Opening this link does not save learning or assess mastery.
Source, access and review notes
Unknown — link only
Check understanding
- Give an example of AI making a mistake (voice assistant mishearing, auto-correct error, wrong recommendation)
- Explain that AI mistakes happen because of gaps or errors in training data
- Describe why AI is not magic — it follows mathematical rules applied to data
“If an AI incorrectly identified a picture of a muffin as a chihuahua, could your child explain why that kind of mistake happens?”
Curriculum record
- Type
- Conceptual
- Subject
- Computing
- Domain
- Artificial Intelligence
- Age range
- Ages 7–9