Findings

Children were asked pre- and post-tests around topics related to machine learning. After their experience with PlushPal, 70% of children who answered the post-survey questions could provide a technical definition of machine learning (one that defines machine learning as a human training a computer to detect something specified). Before PlushPal, only 10% of children were able to provide such a definition.

Children and Data Literacy

Researchers saw through this study that children have different approaches to collecting data samples and highlight some misconceptions, including how machine learning models work. Children had a wide variety of gesture ideas to record, but could use support around how the quantity of sample sizes and similiarities between sample recordings impacts the accuracy of their machine learning models.