This study shows that an optimized children-specific WGAN-GP creative synthesis model can adapt to children's artistic creation, and a 12-week AI-assisted aesthetic education program was associated with improved creativity and aesthetic quality and reduced anxiety, stress and visual fatigue, though the quasi-experimental design and small sample mean results reflect intervention correlation rather than strict causal effect.
The children-specific improved WGAN-GP model produced good image-generation quality and training stability.
Creative thinking scores (TTCT fluency, flexibility, uniqueness) were higher in the AI man-machine collaborative group than in the control group.
Between-group differences in creativity remained significant after controlling for age as a covariate.
Aesthetic literacy scores were higher in the experimental group, with AI-assisted and non-AI works scored separately.
The experimental group had lower anxiety, perceived stress and visual fatigue than the control group.
Physiological and visual indicators after class were more favorable in the experimental group.
Social health expression scores were higher in the experimental group.
The intervention showed a large effect size and good within-group score consistency.
Children's participation in AI creation was strongly positively correlated with creative output quality, but this is cross-sectional correlation.
Children with high baseline anxiety and stress showed the greatest improvement.
The study's design limits causal interpretation, and results are framed as preliminary.
This study built an AI image-generation tool designed specifically for children's art and tested it in a 12-week program with 120 children aged 6 to 14. Children in the experimental group made art together with the AI, while a control group did not. The researchers measured creativity, art skills, anxiety, stress, eye strain, heart rate, and social expression. The AI-assisted group scored higher on creativity and aesthetic measures and reported lower anxiety, stress and visual fatigue. They also had a lower resting heart rate after class and higher social expression scores. The gains held after accounting for age, and children who started with high anxiety and stress appeared to improve the most. The researchers also report that the improvements in creative and aesthetic quality showed up in children's independent work, not only in AI-assisted pieces. The authors stress important limits. The study was quasi-experimental and fairly small, so the results show an association with the program rather than proof that the AI caused the changes. High scores on AI-assisted artwork also should not be read as children's own independent ability. This research suggests that carefully designed AI art tools may be worth studying further for children's well-being, but larger and more rigorous trials are needed.
Chen J, Zhang Y, Liu H. (2026). Generative adversarial network-based creative synthesis and AI art creation in children's art education: effects on health promotion.. Frontiers in public health. https://doi.org/10.3389/fpubh.2026.1945238