Back to The Library
Train the Sorter: How Bias Gets into AI
Created by Argraide
Updated November 7th, 2026
At a glance
- Staff pick
- Picked by the Argraide team
- Play mode
- 👥 Group
- Format
- 🖼️ 2D
- Length
- ⏱ About 20 minutes
- Subject
- Science
- Grade
- 8
- Curriculum
- Alberta
- Curriculum code
- 8CS1
About this activity
A live trail-camera lab. Phones label shots of moose, elk, deer and coyotes; a real closest-match sorter on the big screen learns from the class's labels, aces the summer test, then misses night and deep-snow shots. The class predicts its calls, names where the bias got in, picks and labels the fix, retrains it, and sorts other AIs into supervised, unsupervised and reinforcement learning.
What students learn
- Explain supervised learning: an AI learns from labelled examples (8CS1)
- Predict what a closest-match sorter will say from the examples it learned (8CS1)
- Identify whether bias in an AI was introduced by developers, data sets, algorithms or user input (8CS1)
- Choose training data that fixes a data set bias, and check the fix by retraining (8CS1)
- Compare supervised, unsupervised and reinforcement learning (8CS1)
Reviews
No reviews yet. Be the first to review this activity!
Sign in to review