Despite South Korea's aggressive push toward AI education integration by 2025, early adopter teachers identify resource constraints as the primary barrier, while systemic and cultural factors create an emerging "third-order barrier" that may fundamentally shape the success of technology integration beyond traditional external and internal obstacles.

Objective

The primary goal of this study was to investigate the barriers encountered by five AI education-core teachers—early adopters selected to pioneer AI education integration—as they implemented AI education initiatives across South Korean K-12 schools during the pilot phase (2020-2022). Using Ertmer's framework of technology integration barriers, the researchers sought to identify not only first-order (external/physical) and second-order (internal/perceptual) barriers but also to explore whether additional barriers existed beyond this traditional framework. The study aimed to provide actionable insights for policymakers and educators as South Korea prepares for nationwide AI education implementation.

Methods

The researchers employed a qualitative case study approach involving five male elementary school teachers selected through purposive and peer-recommendation sampling strategies. All participants were designated as AI education-core teachers—a special designation given by the South Korean government to teachers leading AI integration efforts in their districts. Data collection occurred in Fall 2022 through semi-structured virtual interviews lasting 30-60 minutes, conducted in Korean (the participants' and lead author's native language). The interview protocol included questions about professional development experiences, AI education implementation challenges, and observed barriers among colleagues.

Data analysis followed Merriam and Tisdell's qualitative analysis approach, incorporating open coding and thematic analysis based on Ertmer's framework. Both authors independently coded transcripts, categorized barriers according to first-order and second-order classifications, and identified emerging themes outside the framework. The researchers created detailed portraits of each participant to provide context for their experiences and perspectives, enhancing the interpretation and depth of the findings.

Key Findings

The study revealed three primary themes representing distinct barrier types:

Theme 1: Advocating for More Resources (First-Order Barriers)

The participants consistently identified external, resource-related challenges as significant obstacles. These included:

• Training deficiencies: Teachers needed more comprehensive, long-term professional development rather than one-time workshops. Participants advocated for year-long training programs and customized sessions based on teachers' technology competency levels.
• Instructional resources: Schools received insufficient copies of AI textbooks (only 20 copies per school), and existing instructional materials lacked depth for effective content delivery.
• Infrastructure limitations: While schools had basic technology (tablets, Wi-Fi), they lacked appropriate hardware for AI education. Tablets proved unsuitable for machine learning programs that required computers. Additionally, most AI tools were available only in English, creating language barriers, and tools were scattered across multiple platforms rather than unified.
• Teaching hours: Limited instructional time posed challenges, as AI was considered part of software education rather than a standalone subject, with only 17 hours allocated—deemed insufficient for comprehensive AI instruction.

Theme 2: Slight Intrinsic Conflicts (Second-Order Barriers)

Interestingly, the study found relatively minor second-order barriers compared to expectations:

• Resistance to change: Some teachers, particularly senior educators, showed reluctance to learn new technology, though this was minimal among the core teacher participants themselves.
• Teachers' buy-in: While most colleagues recognized AI education's importance, some showed limited interest. Participants addressed this by making training immediately applicable to classroom practice.
• Motivation: Supportive administrators and enthusiastic students/parents served as strong motivators. Participants credited "forward-thinking" principals and positive feedback from parents as driving forces.
• Teachers' confidence: Participants anticipated initial confidence issues among colleagues but believed that with proper training and based on previous successful software education integration, teachers would become more confident over time.

Theme 3: Reflecting on Roles Within AI Integration Policy (Potential Third-Order Barriers)

The study identified barriers beyond Ertmer's traditional framework, related to broader educational systems and sociocultural contexts:

• Questioning the status quo: Participants expressed concerns about curriculum limitations, wanting AI education to extend beyond basic technical skills to include problem-solving, critical thinking, and addressing social issues using AI. They desired AI literacy that empowered students as "creative co-creators" rather than mere users.
• Confined by the system: South Korea's centralized, top-down education system limited teachers' autonomy. As one participant noted, "We do not have the autonomy to change anything." The collectivist culture promoted rapid technology adoption without extensive critical discussion, sometimes prioritizing trendy topics over meaningful integration. This systematic constraint created barriers that teachers recognized but felt powerless to address.

The researchers proposed three potential relationships between these barrier types: a government-led approach (where first-order barriers dominate), a collaborative approach (where third-order barriers equal first-order barriers in significance), and a teacher-led approach (where third-order barriers become most prominent and influential).

Implications

This research makes several significant contributions to AI education integration:

1. Tailored strategies needed: Despite South Korea's advanced technology integration history, barriers persist that require customized policies addressing specific contextual challenges rather than one-size-fits-all solutions.
2. Third-order barriers framework: The identification of systemic and sociocultural barriers that transcend individual or institutional factors adds a new dimension to technology integration research, suggesting that successful implementation requires addressing broader educational system structures and cultural norms.
3. AI literacy reconceptualization: The findings emphasize that AI education must evolve beyond technical skill development to encompass comprehensive AI literacy including critical thinking, computational thinking, problem-solving, ethics, and cultural implications.
4. Teacher voice integration: The study highlights the importance of incorporating practitioner perspectives into policy development, particularly in centralized education systems where teachers may feel constrained by top-down mandates.

Limitations

The researchers acknowledge several important limitations:

• Demographic constraints: The study focused exclusively on five male elementary school teachers, limiting representation of diverse educator perspectives across gender, school levels, and experience ranges.
• Temporal scope: Data collection occurred during the early pilot phase (Fall 2022), before full curriculum implementation in 2025, meaning findings represent only initial-stage challenges that may evolve.
• Methodological boundaries: The qualitative case study approach, while providing deep insights, limits generalizability beyond similar educational and cultural contexts.
• Sample size: Five participants, though information-rich as core teachers who also trained colleagues, represent a limited sample that may not capture the full spectrum of barriers across South Korea's diverse educational landscape.

Future Directions

The researchers recommend several avenues for continued investigation:

1. Barrier relationship exploration: Further research should investigate how first-order, second-order, and third-order barriers interact dynamically, including their relative magnitude and influence on technology integration success.
2. Expanded demographic sampling: Future studies should include diverse participants across gender, school levels (middle and high school), geographic regions, and varying AI competency levels to validate and generalize findings.
3. Longitudinal investigation: As South Korea implements AI education nationwide post-2025, ongoing research should track how barriers evolve, whether first-order barriers diminish as resources increase, and whether third-order barriers become more prominent.
4. Comparative international studies: Research should examine AI integration barriers in countries with similar centralized education systems or cultural contexts to identify common challenges and effective strategies.
5. AI competency framework validation: Studies should explore how emerging frameworks (such as UNESCO's AI competency guidelines for teachers and students) can address identified barriers and support effective integration.
6. Teacher autonomy models: Investigation into how different governance structures (government-led, collaborative, or teacher-led approaches) impact barrier prevalence and integration success could inform policy development.

Title and Authors: "Barriers to Integrating Artificial Intelligence Education: Implications from Five Early Adopters in South Korea" by Wanju Huang (Purdue University) and Wonjin Yu (University of Alabama)

Published On: September 29, 2025

Published By: Computer Science Education (Taylor & Francis Group)