Exploring Factors Influencing Preschool Teachers' Behavioral Intention to Use AI Technologies in Early Childhood Settings

Synthesis: Duan, Shan, and Gong examine what shapes preschool teachers' willingness to adopt AI in early childhood classrooms. Grounded in an extended Technology Acceptance Model (TAM), the study uses structural equation modeling to test how perceived usefulness (PU), perceived ease of use (PEOU), subjective norm (SN), AI anxiety (AIA), and AI Self-Efficacy (AISE) influence preschool teachers' behavioral intention to use AI.

Key Findings

• Perceived usefulness and ease of use remain central drivers, consistent with the core TAM.
• AI self-efficacy emerges as a meaningful predictor of intention, underscoring confidence-building as a lever for early-childhood AI adoption.
• AI anxiety acts as a deterrent, and subjective norm (social influence from colleagues and leadership) shapes intention.
• The extended-TAM model is validated in the distinctive early-childhood context, where developmental appropriateness and teacher comfort matter for AI integration.

What this means for practice

• Administrators. Lower AI anxiety and raise Self-Efficacy with hands-on, low-stakes training, since AI self-efficacy predicted intention while AI anxiety deterred it.
• Teacher educators. Give pre-service preschool teachers developmentally grounded experience with teacher-facing AI — lesson preparation, administration, materials — rather than child-facing tools.
• Designers. Compete on perceived usefulness and ease of use for tools aimed at teachers' professional work; the study deliberately excluded AI applications directed at children.
• Administrators. Require age-appropriateness review before any tool enters an early childhood setting, since adoption intent here is shaped by perceived value and social pressure rather than by developmental fit.

Limitations

• The 270 valid responses (33 questionnaires excluded) came from a single online survey run in September and October 2025, mostly in the relatively developed regions of Shanghai, Zhejiang, and Guangdong.
• The sample is heavily female (250 women, 93%; 20 men, 7%) and 81.9% (n = 221) already had experience with AI, so gender and education could not be modeled as moderators.
• All six constructs — perceived usefulness, ease of use, subjective norm, AI anxiety, AI self-efficacy, and behavioral intention — were self-reported Likert items, leaving the relationships open to common-method effects.

Citation

Duan, Z., Shan, Y., & Gong, Y. (2026). Exploring factors influencing preschool teachers' behavioral intention to use AI technologies in early childhood settings. Computers and Education: Artificial Intelligence, 10, 100589.