At Futurebytes, we are dedicated to shaping the innovators of tomorrow. With a mission to make learning accessible and engaging, we’re excited to inspire young minds this summer and beyond. See you this summer at Futurebytes—where curiosity meets creativity!
An interactive artificial intelligence camp should let children investigate a system rather than watch it produce impressive answers. A small image-classification experiment offers a clear starting point. Students can choose examples, test new objects, and explore what changes when the conditions around a camera change.
Start with a classification game
Before opening software, sort picture cards into two categories and explain the rule. Then introduce a confusing example. Children can discuss why a category is useful, what makes an example ambiguous, and whether a person might reasonably disagree about its label.
This activity prepares learners to think about a classifier: a model that assigns an input to a category. Google's Teachable Machine provides a visual way to gather examples, train a model, and test classifications. [1] An instructor can use ordinary objects so the investigation does not require identifiable student images.
Change the conditions and observe the prediction
Consider training with two kinds of classroom objects, then testing them against a new background. Move one object farther from the camera or change its angle. The aim is to discover when the output changes and whether the model relied on an unhelpful visual clue.
A model's output score should not be treated as a guarantee that the answer is correct. Keep a simple record of inputs and results, then examine mistakes. Children can propose a better collection of examples and compare the next version under the same test conditions.
Connect the activity with responsible choices
Discuss where a similar system could be useful and what might happen when it is wrong. An error sorting toy pictures has different consequences from an error affecting a person. AI4K12 includes societal impact as an essential part of AI education. [2]
Futurebytes' AI Developer program includes guided AI and machine learning projects. [3] Families in Santa Clara, Cupertino, Los Altos, and nearby San Jose should check the tools and learning level in the current camp listing. A good takeaway is the ability to explain an AI experiment and identify a situation where human judgment is needed.
Resources
[1] Google —Teachable Machine workflow
.png){kind=logo}
