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
Compare teaching approaches
Teaching approaches
Gakuva’s recommended starting approach is selected. You can still compare every option.
1 approach
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.
Materials: Two short recommendation cards or screenshot printouts; Pencil for compare notes. 1. Say the goal of AI Mistakes and Limitations in plain words once. 2. Adult shows one AI suggestion and asks what data might have caused it. 3. Child reviews three AI examples and names one input clue and one possible bias or limit each. 4. Fresh check: one fresh AI example with a limit. Record and stop. Done: three AI examples with an input clue and a limit note. Proof: parent note of examples and results. Stop: End after the fresh check; keep optional challenge for another day.
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