Exploring Middle School Students’ Interactions and Perceptions of Using Generative AI in STEAM Learning Environments

Editorial reading guide · Independence

When AI helps with a project, what decisions and checks should remain my child’s?

Offers a concrete example of high satisfaction and trust coexisting with limited verification and little sense of accomplishment. Students mainly retrieved information and copied instructions, making the article useful for discussing which thinking a learner should keep doing when an answer is readily available.

Study context

12 students in Zhejiang, China, working in six groups during two 120-minute STEAM modules

Age / stage: 13–14 years; eighth grade

Read with three questions

1. Where did the students request information instead of evaluating or revising an AI suggestion?
2. What could a prompt or adult question change so the learner explains a choice before receiving more help?
3. How do the students’ satisfaction and limited accomplishment complicate the idea that an enjoyable tool supports agency?

Keep the limits in view

The small, gender-imbalanced convenience sample, off-campus setting, short exposure, and absent comparison group preclude effectiveness claims. Module order and different instructors were confounded; researchers also supplied prompting help and conducted the interviews. Classroom/home transfer and sustained independence were not measured.