Artificial intelligence curriculum for primary and secondary schools
AI in Informatics at the 1st level
Ju and Pi on shopping
Bias
Lessons 07
Rewiring
In order for robots to know what they are looking at, we have to show them lots and lots of things and name them. But when we show robots the wrong examples or give them too few, they sometimes get confused. This time the children are in for an adventure in the supermarket!
How the lesson works
The lesson begins with a story in which the robots Ju and Pi go shopping with the task of bringing apples and pineapples. Although they have expert books full of information, they are unable to find the fruit because they expect to find apple trees (trees) or a bush full of pineapples. The teacher leads a discussion about how humans form ideas based on a variety of experiences and examples, while machines learn only from the data we provide. Children reflect on the different shapes an apple or pineapple might take and how difficult it might be to recognize them if the model has not been trained on a sufficiently diverse set of images. In a hands-on activity, students try to train their own machine learning model using Teachable Machine - they prepare a training set and then test how well the model recognises different objects. This is followed by a discussion of how bias arises and how it can affect the decision making of AI systems, for example in face recognition or in recommendation algorithms. The lesson ends with a reflection where children share their observations and discuss how machine learning models could be improved by providing better and more diverse data.
Information about lessons
Subsidies and years
45-90 minutes, grades 3-5 Elementary school
Aids
Educator: projection equipment and presentation to project, camera
Students: creative activity supplies, equipment for groups
Building stones
Bias
What pupils learn
If we prepare bad data from which computers learn to recognize things, then these computers also recognize them badly.
Why is it taught
Based on their understanding of the concept of bias, they critically assess the functioning of the AI system.
How do we know if they've learned
They will prepare data to train the machine learning model and test whether the model can correctly recognize different things with this data.
Outputs of the RVP
Computer Science:
Data, Information, and Modeling:
I-5-1-01 gives examples of data surrounding him/her that can help him/her make better decisions; expresses answers based on data
I-5-1-03 reads information from a given model
Bloom's taxonomy
Memorization: students explain the concept of bias.
Application: prepares data for training a machine learning model.
Rate: Evaluate the performance of the machine learning model.
Digital competences
Information and communication
Bloom's taxonomy
Application: Students prepare data for training a machine learning model.
Analysis: They analyze how the quality and variety of the training data affect the accuracy of the model.
Evaluation: They will evaluate the performance of the machine learning model.
Five Big Ideas
2-A-IV Representation (feature vectors)
2-C-II Inference (inference algorithms)
3-A-VI The nature of learning (learning from experience)
3-C-III Datasets (bias)
