Creating a Playful AI/ML Learning Experience for Indonesian Families

KueAI is an informal, family-centered learning experience introducing foundational AI concepts through a playful competition embedded in Indonesia’s traditional Independence Day games and celebrations.

• Student-parent teams build snack models (klepon and onde-onde) with craft supplies, then train image classifiers using Google’s Teachable Machine.

• The activity invites participants to build mental models of AI concepts like perception, representation, and machine learning without relying on technical instruction.

• Take-home booklets and a companion chatbot extend the learning, offering games, reflections, and simplified lessons in Bahasa Indonesia — each building on the hands-on activity to help families connect abstract AI concepts to their hands-on experience.

The Competition

Phase 1

Use only the provided craft materials to create representations of Klepon and without references.

Some ambiguous or seemingly irrelevant materials are intentionally included so that, in deciding what to use, participants simulate the process of “feature selection” in machine learning.

Most participants will rely on prior knowledge, while others must make creative guesses—simulating “representation under uncertainty.” Their models reflect assumed relevant features, enacting analog feature engineering and probabilistic reasoning.

Phase 2

Train a Teachable Machine image classifier to recognize the klepon and onde-onde using the crafted models.

After initial training, the actual snacks are revealed, letting teams test how well their models generalize. Successes and failures prompt evaluation of how data quality, variation, and labeling influence an AI system’s accuracy.

Phase 3

Test and refine models with real Klepon & Onde-Onde. Team with the most accurate model win!

Learning at Home

The competition is designed to spark curiosity and serve as a conceptual foothold for continued learning at home. After the event, families receive a set of illustrated booklets in plain Bahasa Indonesia, each offering simple lessons and activities on foundational AI topics: What is AI? How do machines learn? How do machines create? The booklets are currently in development and draw from existing AI literacy frameworks, but will be refined in collaboration with Indonesian educators to ensure content relevance and accessibility.

Each booklet also includes a QR code linking to the KueAI GPT, a custom chatbot that helps families revisit concepts through examples from the hands-on activity or play an AI guessing game at their own pace. Post-event engagement with the bot serves as an informal signal that curiosity has taken root.

If you are an Indonesian educator or AI expert and interested in designing exploratory, hands-on, low-tech AI learning experiences and content—especially for communities often left out of the conversation—please reach out! I'd love to connect and collaborate.