Teaching Google's Teachable Machine to Elementary School Students

My mentor teacher, Judith, and I led our 4th grade students through a lesson using Google’s Teachable Machine. We based our lesson on Ready AI’s lesson, modifying it to focus on colors and artistic styles rather than fish identification. Students used images of famous artists’ works, trained the model to recognize distinctive styles (Picasso, Van Gogh, etc.), and deliberately set aside some images as a testing set that the model would never see during training.

Our lesson addresses NYS Computer Science and Digital Fluency Standards, particularly 4-6.IC.5 (explaining how computer systems play a role in human decision-making) and 4-6.CT.1 (developing a computational model showing changes in output when inputs change). Students immediately noticed how the model’s confidence levels fluctuated when running inference, leading seamlessly into subsequent class discussions about AI reliability and bias.

The most valuable learning moment came when testing the “never seen” images, addressing standard 4-6.CT.3 (visualizing data to highlight relationships). Students were curious why certain artistic styles were easier for the machine to identify than others, and developed hypotheses about feature recognition. Student enthusiasm was contagious - students were excited for class knowing they were going to train an AI model, and they have already asked when they can build their next AI project!

For more information about the NYS Computer Science and Digital Fluency Standards for grades 4-6 please see here.