FOSTERING CRITICAL THINKING TOWARD AI GENERATED CONTENT IN MIDDLE SCHOOL STUDENTS: A QUASI-EXPERIMENTAL EVALUATION OF A DIGITAL CITIZENSHIP INTERVENTION

<doi>10.24250/jpe/SI/2026/SI/</doi>

Abstract

Generative artificial intelligence tools have become
embedded in the everyday academic lives of adolescents,
yet the capacity of middle school students to critically
appraise
AI-generated
content
remains
largely
unaddressed by current digital citizenship curricula. This
study evaluated a five-session educational intervention
designed to develop critical thinking toward AI-generated
content among seventh-grade students, grounded in
psychological inoculation theory and AI literacy
competency models. A quasi-experimental pretest-posttest
design with a non-equivalent control group was applied to
275 seventh-grade students (experimental group, n = 138;
control group, n = 137) recruited from parallel classes of
two middle school in Arad, Romania. Critical thinking
toward AI-generated content was measured with an
original 20-item Likert scale, and discrimination ability
was assessed through a practical task requiring students to
classify ten text and image stimuli as AI-generated or
human-produced. Following the intervention, the
experimental group showed a substantial increase in
critical thinking scores, from M = 54.25 (SD = 8.12) to M
= 68.90 (SD = 7.65), t(39) = -9.85, p < .001, d = 1.86, and
in discrimination accuracy, from 51.25% to 68.50%, t(39)
= -8.14, p < .001, d = 1.65. The control group showed no
significant change on either measure. Critical thinking
scores and discrimination accuracy correlated moderately
to strongly at post-test, r = .58, p < .001. These findings
support the feasibility of short, curriculum-embedded
interventions grounded in inoculation theory for cultivating
AI literacy at the middle school level, and they extend the
construct of digital citizenship to include critical
engagement with AI-generated content as a measurable,
educable competency.