GenAI is transforming K-12 classrooms by creating a new collaborative division of labor where teachers evolve from knowledge transmitters to instructional decision-makers and ethical guides while AI handles routine tasks.
Objective: This systematic review aimed to examine how Generative Artificial Intelligence (GenAI) reconfigures teachers' instructional work in K-12 classrooms and to develop a conceptual framework for understanding human-AI collaboration across the three fundamental stages of teaching: planning, implementation, and assessment. The researchers sought to understand how these collaborative mechanisms influence teacher role transformation and professional identity reconstruction in the era of AI-enhanced education.
Methods: The researchers conducted a comprehensive systematic literature review following established methodological principles. They searched three major databases (Web of Science, Education Resources Information Center, and APA PsycInfo) for empirical studies published between November 30, 2022 (ChatGPT's release date) and December 31, 2024. Using Boolean search terms related to generative AI and K-12 education, they initially identified 1,277 studies. After applying rigorous inclusion criteria - peer-reviewed empirical research focusing on GenAI and teachers' work in K-12 classroom teaching written in English - they systematically screened and analyzed 42 studies. The analysis involved four steps: extracting study characteristics, categorizing findings into classroom applications, teachers' work, and application outcomes, organizing results according to Shulman's three-stage instructional framework, and ensuring reliability through independent coding by multiple researchers.
Key Findings: The review revealed that teacher-GenAI collaboration follows a division of labor based on comparative advantages, with three distinct interaction patterns emerging across instructional stages. During planning, a human-centric model predominates where teachers set instructional frameworks and objectives while GenAI provides content support and generates instructional designs. In implementation, an AI-centric model emerges where GenAI takes primary responsibility for delivering content and providing cognitive scaffolding, while teachers focus on supervision, emotional support, and ethical guidance. The assessment stage demonstrates a human-AI symbiotic model where both parties contribute collaboratively to data analysis and decision-making.
The study identified three types of transformation in teachers' work: replacement of routine and highly structured tasks (such as resource collection and basic grading), reinforcement of judgment-intensive responsibilities in complex contexts (including instructional decision-making and cognitive facilitation), and expansion into new AI-related duties (such as AI tool evaluation, content supervision, and digital literacy guidance). Teachers' roles are increasingly centered on being instructional decision-makers, cognitive facilitators, social-emotional supporters, AI tool evaluators, and digital literacy guides.
Furthermore, the research found that classroom authority is being reconfigured from knowledge exclusivity toward composite professional authority grounded in instructional decision-making, humanistic care, and value leadership. Teachers retain ultimate control over decision-making, regulation, and value-laden dimensions of instruction, maintaining a predominantly human-centered interaction logic even as AI participation increases.
Implications: These findings provide crucial insights for educational stakeholders navigating the integration of AI in K-12 settings. The research demonstrates that rather than replacing teachers, GenAI is reshaping their professional practice by automating routine tasks while amplifying their roles as ethical guides, emotional supporters, and instructional decision-makers. This has significant implications for teacher education programs, which must evolve to include AI literacy, prompt engineering skills, and competencies in human-AI collaboration. Educational policymakers can use these insights to develop frameworks that maximize GenAI's potential while preserving the humanistic dimensions of teaching. The study also suggests that successful AI integration requires viewing teachers as central agents who orchestrate collaboration rather than passive recipients of technological change.
Limitations: The study acknowledges several important limitations. The research covers the early adoption phase of GenAI in education, potentially missing long-term impacts as the technology matures. The literature is predominantly concentrated in humanities and natural sciences, limiting generalizability across all subject areas. Additionally, the review's timeframe, while comprehensive for the period following ChatGPT's release, may not capture the full spectrum of evolving AI capabilities and their educational applications. The focus on teacher roles, while valuable, does not systematically address the broader teacher-AI-student triadic relationship dynamics.
Future Directions: The researchers propose several avenues for future investigation. Technology developers should create more education-specific AI tools that align with pedagogical principles and address teachers' practical needs, particularly in areas like emotional responsiveness and subject-specific expertise. Teacher education researchers should continuously update competency frameworks to include prompt engineering, AI ethics, and affective regulation skills. Future studies should explore GenAI applications across diverse disciplines beyond humanities and sciences, examine long-term impacts in authentic classroom settings, and investigate the complete teacher-AI-student relationship dynamics. The researchers also emphasize the need for larger-scale studies that can capture the nuanced ways different subject areas and educational contexts shape human-AI collaboration patterns.
Title and Authors: "Reconfiguring Teachers' Instructional Work in the Era of Generative Artificial Intelligence: A Systematic Review of Human–AI Collaboration in K–12 Classroom Teaching" by Wanya Liang, Sofia Rose Hernandez, Simon Tzeng, Ke Guo, Yi Guan, and Longhai Xiao.
Published On: 2025 (based on the systematic review methodology covering studies through December 31, 2024)
Published By: This appears to be a research manuscript from Zhejiang University's College of Education and collaborating institutions, though the specific journal publication details are not clearly indicated in the provided document.