Exploring AI Problem Formulation with Children Via Teachable Machines

Emphasizing problem formulation in AI literacy activities with children isvital, yet we lack empirical studies on their structure and affordances. Wepropose that participatory design involving teachable machines facilitatesproblem formulation activities. To test this, we integrated problem reductionheuristics into storyboarding and invited a university-based intergenerationaldesign team of 10 children (ages 8-13) and 9 adults to co-design a teachablemachine. We find that children draw from personal experiences when formulatingAI problems; they assume voice and video capabilities, explore diverse machinelearning approaches, and plan for error handling. Their ideas promote humaninvolvement in AI, though some are drawn to more autonomous systems. Theirdesigns prioritize values like capability, logic, helpfulness, responsibility,and obedience, and a preference for a comfortable life, family security, innerharmony, and excitement as end-states. We conclude by discussing how theseresults can inform the design of future participatory AI activities.