Objective: The main goal of this systematic review was to investigate the current state of research on generative artificial intelligence (GenAI) tools in science education by analyzing 41 peer-reviewed articles from the Web of Science database. The study aimed to examine how GenAI is being integrated into science education across different sub-disciplines, topics, application levels, participant profiles, and methodological approaches, while identifying gaps and future research directions.
Methods: The researchers conducted a systematic literature review following PRISMA 2020 guidelines, screening the Web of Science database for studies published from 2022 onwards using two search clusters: ChatGPT-focused searches and broader GenAI searches. They employed the PICRAT (Passive, Interactive, Creative / Replacement, Amplification, Transformation) technology integration model as their theoretical framework to analyze how GenAI tools are being used in science education. A comprehensive coding scheme was developed to analyze various parameters including science sub-disciplines, intended use, student-centeredness, application levels, instructional components, participant profiles, research methods, and technology integration models. Inter-coder reliability was established at 0.95, and both deductive and inductive coding approaches were used for data analysis.
Key Findings: The review revealed several critical insights about GenAI use in science education:
- Disciplinary focus: Chemistry education dominated with 12 studies (29%), followed by physics (17%) and biology (5%), while 41% of studies were not field-specific.
- Intended use: The majority of studies (51%) focused on evaluating GenAI tools' performance rather than implementing them for actual teaching enhancement, with only one study (2%) aimed at enriching science teaching.
- Student-centeredness: Only 12 out of 41 studies involved students using GenAI tools, with just 5 studies allowing student-centered use without guidance.
- Application level: According to the PICRAT framework, 14 studies used GenAI at a replacement level, indicating basic substitution rather than transformative integration.
- Instructional components: Assessment was the most studied component (55%), while other aspects like curriculum design and instructional strategies received minimal attention.
- Research methods: Qualitative methods were most common (41%), followed by quantitative (24%) and mixed methods (17%).
- Geographic distribution: The USA led with 26% of studies, followed by Australia (16%), with limited representation from non-English speaking countries.
Implications: This research reveals that GenAI integration in science education is still in its exploratory phase, primarily focused on tool evaluation rather than meaningful pedagogical integration. The findings suggest an urgent need for interdisciplinary partnerships between technology and science educators to develop more comprehensive approaches to GenAI integration. The study highlights the importance of moving beyond simple replacement-level uses toward more transformative applications that can genuinely enhance science learning. The research also emphasizes the need for more student-centered approaches and comprehensive teacher training programs to effectively integrate GenAI tools into science instruction.
Limitations: The study was limited to English-language publications from the Web of Science database, potentially missing research from non-English speaking countries and other databases. The rapid evolution of GenAI technology means that some emerging tools or terminologies may not have been captured in the search strategy. The review was also constrained by the timeframe (2022-2024) and the infancy of the field, which resulted in limited empirical studies with actual classroom implementations. Additionally, the focus on only one theoretical framework (PICRAT) may have limited the analysis of technology integration from other perspectives.
Future Directions: The researchers recommend several key areas for future investigation: developing more comprehensive pedagogical approaches that move beyond tool evaluation to actual classroom implementation; conducting longitudinal studies to examine the sustained impact of GenAI on student learning outcomes; exploring GenAI applications across diverse educational settings and cultural contexts; investigating how GenAI can enhance specific science education practices like argumentation, scientific modeling, and nature of science instruction; developing teacher training programs specifically designed for GenAI integration in science education; and addressing equity issues related to access and cultural compatibility of GenAI tools across different regions and populations.
Title and Authors: "Generative AI as the New Frontier in Science Education: A Systematic Review of Web of Science Articles" by Sevgi Aydin-Günbatar, Alper Durukan, and Mustafa Serkan Günbatar.
Published On: August 2025
Published By: Science & Education