Generative artificial intelligence (GenAI) has rapidly entered science education, but its role in supporting scientific inquiry and science practices remains unclear. This scoping review mapped empirical studies on how GenAI has been used to support inquiry- and practice-oriented work in science education. Guided by JBI scoping review principles and PRISMA-ScR, the review used a reproducible open-data search route through OpenAlex and Crossref, supplemented by targeted manual discovery. Records published from 30 November 2022 onward were screened against predefined criteria, resulting in 14 included empirical journal articles. The reviewed studies clustered around three main pedagogical roles: learner-facing inquiry partners, teacher-facing design and co-design tools, and assessment or feedback agents. Investigation planning, explanation, and inquiry scaffolding were more common than data analysis, argumentation, and explicit epistemic reflection. Overall, the literature suggests that GenAI is currently used more often to structure, prompt, or evaluate inquiry than to support deeper disciplinary participation. Productive use depended heavily on teacher mediation, prompt quality, and learners' critical oversight. The review identifies both practical opportunities and persistent epistemic gaps for future research.
This study examined the relationship between science teachers’ effective science teaching levels and their entrepreneurship education levels. A descriptive, correlational survey design was employed. The sample consisted of 94 science teachers working in middle schools in Kırıkkale province during the 2023–2024 academic year. Data were collected using the Effective Science Teaching Scale and the Entrepreneurship Scale for University Students. Descriptive statistics, independent samples t-tests, one-way ANOVA with LSD post hoc comparisons, and Pearson correlation analysis were conducted. The findings indicated that science teachers demonstrated high levels of both effective science teaching and entrepreneurship education. Significant differences were found in effective science teaching according to professional seniority and gender. Additionally, entrepreneurship education levels differed significantly by educational background and gender. Correlation analysis revealed a statistically significant and positive relationship between effective science teaching and entrepreneurship education levels.
Sustainable development is development that meets the needs of the present without compromising the ability of future generations to meet their own needs. In modern usage it generally refers to a state in which the environment, economy, and society will continue to exist over a long period of timeThis study explored the perceptions and experiences of primary school candidate teachers trained in Sustainable Development Goals (SDGs) regarding real-world issues. Using a phenomenological approach with focus groups and semi-structured interviews, this study aims to understand the impact of SDG-related education on these candidates. Data analysis involved thematic and content analysis, revealing that teacher candidates saw a direct link between SDGs and global issues, often sharing insights based on personal experiences. They highlighted the importance of integrating the SDGs into the curriculum, suggesting both compulsory courses and projects. The candidates also prioritized specific SDGs when designing activities that were personally significant. This small-scale study aims to inform future research on SDGs and real-world problem-solving in education.
This review provides an overview of how GAI assists in chemistry learning and teaching. Aspects such as 21st century skills and the tutoring ability of GAI were examined from the learning point of view, while also covering pedagogical aspects such as Technological Pedagogical Content Knowledge (TPACK), visualization and representation levels of Chemistry, textual aspects, and problem-based learning (PBL). The GAI seemed to elicit students’ 21st century skills and, to a lesser extent, display a tutoring ability, depending on whether it was a free or paid version. Furthermore, the GAI requires objectivity and sufficient TPACK owing to its hallucinations and inconsistencies in definitions, diagram interpretation, and image generation, among others. Furthermore, the chatbots seemed to struggle with representation levels, generating responses with macro-level definitions for concepts requiring definitions at the sub-micro level. These inconsistencies and hallucinations, without meticulous verification, can lead to misconceptions. Moreover, social inequalities may negatively impact meaningful learning, as paid chatbots perform better than free versions in generating and interpreting chemistry images, among other abilities.
In this study, a four-tier concept test was developed to determine the conceptual understanding and misconceptions of 8th-grade middle school students regarding the topic of matter. The Matter Concept Test (MCT) was prepared with 16 items, each consisting of four steps, and applied to 175 eighth-grade students studying in Eskişehir. SPSS, Excel, and Factor statistical programs were used to analyze the data. According to the results of the exploratory factor analysis, the test's KMO value was found to be .652, exhibiting a four-factor structure with eigenvalues above one and explaining 42.4% of the total variance. All factor loadings were above .30, supporting the construct validity of the items. In the reliability analysis, the KR-20 coefficient for the scientific knowledge score was 0.816, and the KR-20 coefficient for the misconception score was 0.743. Both values being above .70 indicate that the test is reliable. Furthermore, the false positive average was calculated to be 5.2%, and the false negative average was calculated to be 3.7%. Both ratios being below the 10% threshold specified in the literature support the validity of the test. When the item difficulty and discrimination indices were examined, it was seen that the test consisted of items of medium difficulty and high discrimination. The findings showed that students had clear misconceptions about pure matters, atoms, mixtures and the separation of mixtures. This finding supports that the developed test can identify conceptual errors across multiple dimensions, including content knowledge, reasoning and confidence levels. In conclusion, the developed four-tier concept test was evaluated as a measurement tool that reliably and validly reveals both students' scientific knowledge and their misconceptions.
The rapid diffusion of generative artificial intelligence (AI) tools into educational contexts has fundamentally transformed how students approach homework, academic writing, and independent learning tasks. Whilst AI-assisted homework tools promise efficiency, personalization, and immediate feedback, there remains some debate over their implications for academic performance and learning quality. The present study proffers a thorough synthesis of empirical evidence, examining how students' academic performance differs when using AI homework tools compared to traditional homework methods. The review draws on experimental, quasi-experimental, and observational research conducted across secondary and higher education contexts. The findings of the study indicate that AI homework tools are associated with significantly higher grades and writing scores in most controlled comparisons, particularly in language learning contexts, with effect sizes ranging from medium to large. However, the evidence also reveals important trade-offs, including reduced knowledge retention, lower originality, and diminished critical thinking in some settings. The synthesis demonstrates that AI tools primarily optimize output quality rather than learning processes, and that their effectiveness is highly conditional on task characteristics, assessment timing, implementation fidelity, and learner characteristics.
Argumentation has become a cornerstone of science education research, essential for fostering evidence-based thinking, scientific reasoning, and scientific literacy. This study employs bibliometric methods to examine global publication trends, intellectual structures, and thematic transformations in argumentation-focused research between 2001 and 2025. A final dataset of 474 peer-reviewed articles, retrieved from the Web of Science Core Collection after applying stringent inclusion and exclusion criteria, was analyzed using the R-based Bibliometrix package. To ensure a holistic interpretation, author keywords were standardized and categorized into overarching themes, allowing the field’s conceptual landscape to be mapped in a more integrated manner rather than through fragmented indicators. The results indicate that argumentation studies emerged between 2001 and 2010, experienced rapid growth from 2011 to 2020, and approached a maturation phase by 2025. Early research focused mainly on cognitive argument structures, while later work increasingly engaged with socioscientific issues, epistemic practices, and classroom discourse. During this process, classical models were recontextualized within contemporary pedagogical settings and underwent terminological transformation. By illustrating the transformation of the science education argumentation literature from a pedagogical tool to a foundational epistemic framework, this study offers an empirically grounded perspective that may inform future research directions in the field.
The aim of this study is to examine the role of artificial intelligence (AI) in physics teaching and learning between 2015 and 2025 through a systematic literature review. The research process was conducted in accordance with PRISMA 2020 guidelines; a total of 11,208 records were identified through searches in the Web of Science, Scopus, and ERIC databases. After removing duplicate records and applying the inclusion criteria, 40 studies were included in the final analysis. The findings indicate a marked increase in AI-supported physics education research, particularly after 2023. Most of the studies employed a mixed-methods design, with undergraduate students predominantly selected as the sample group. The results of the content analysis reveal that AI applications are most frequently concentrated in mechanics topics, followed by electromagnetism and thermodynamics. A significant proportion of the research focuses on examining the performance of generative AI systems in problem-solving, automated assessment, and personalized feedback processes. However, teacher-focused studies and long-term analyses of pedagogical impact appear to be limited. In conclusion, the field of AI-supported physics education is undergoing rapid development; however, more comprehensive research is needed in terms of methodological diversity, sample balance, and pedagogical depth.
The purpose of this study was to develop a valid and reliable measurement tool that can be used to identify pre-service teachers' misconceptions about electrochemistry. The Electrochemistry Concept Test (ECT) developed for this purpose consists of a total of 12 questions, each containing four tiers. The Electrochemistry Concept Test (ECT) was applied to a total of 307 science teacher candidates from 10 different universities in seven regions of Türkiye. The data were analyzed using Excel, SPSS, and Factor programs. A four-factor structure with 12 questions was determined using EFA, and the KMO value of the test was calculated as 0,682, indicating that this four-factor structure explained 48% of the total variance. As a result of the reliability analyses, the KR-20 reliability coefficient for the scientific knowledge score obtained from correct answers was calculated as .819, and the KR-20 reliability coefficient for the misconception score obtained from wrong answers was calculated as 0,702. The item analyses revealed that the difficulty and discriminative indices of the test were at an intermediate level. Positive and negative wrong values were calculated for content validity. The validity and reliability analyses confirm that the developed test is a valid and reliable measurement tool. As a result of the analysis of the data collected in the study, it was determined that the pre-service teachers had misconceptions above 10% for each question related to electrochemistry - activity, reduction–oxidation, galvanic cell reactions, and cell potential - and that their scientific knowledge was not at a sufficient level. As a result of the study, a total of 37 misconceptions related to the topic of electrochemistry were identified, among which 9 misconceptions with a prevalence of over 10% were determined, thereby contributing new misconception statements to existing literature. In conclusion, a valid and reliable measurement tool has been developed that researchers can use to determine pre-service teachers' misconceptions and achievement levels in electrochemistry.
Environmental education is the solution to the environmental problems caused by technological advancements. Technological advancements make human life easier in many aspect. Today, artificial intelligence is one of the current technological advancements. The purpose of this research is to determine the general trend in studies on the use of artificial intelligence in environmental education, identify gaps in the relevant literature, and explain how current developments in educational technologies impact environmental education. This research is a systematic review. In this research, studies on the use of artificial intelligence in environmental education were systematically compiled to reveal how new developments in artificial intelligence affect the environmental education process. Seventeen studies were included in the review process. This research has two main conclusions. First, artificial intelligence in environmental education is a relatively new and current topic in literature. Second, the use of artificial intelligence tools causes water waste and carbon emissions. Therefore, the conscious use of artificial intelligence is essential for both teachers and students to be aware of it.
This study investigated the impact of school setting, gender, and age on Namibian primary students’ learning achievement following an inquiry‑based science fieldwork (IBSF) intervention and examined how attitude towards IBSF varied by these factors. A mixed-methods approach was used with 100 seventh-grade students from two socio-demographically different schools. From this group, 20 students were purposively selected for semi-structured interviews. Thus, multiple linear regression, exploratory factor analysis, multivariate analysis of variance, and Pearson correlation coefficients were conducted. Our quantitative results showed that school setting—but not gender or age—significantly influenced post‑test learning achievement, underlining the critical role of the school environment in meeting learning outcomes. In contrast, school setting had no significant effect on students’ attitudes toward IBSF. Gender differences emerged in attitudes: boys reported more positive views of IBSF’s importance and greater eagerness to participate in future fieldwork, whereas girls expressed some reluctance despite strong performance. Content analysis of qualitative data confirmed that most students valued IBSF and were eager to engage again. These findings suggest that introducing IBSF at the primary level enhances students' understanding of science and its relevance to their futures. The study recommends that interventions also address gender stereotypes to promote equity and sustained interest in science education.
Pestalozzi emphasised that nature and life outside school are valuable teachers for children, stating, ‘When the birds are chirping beautifully and a worm is crawling on a leaf, stop your language studies immediately. Know that the bird and the worm teach the child better and more.’ All educational activities that take place outside of school throughout an individual's life are defined as ‘out-of-school education,’ and the learning occuring during this process is defined as ‘out-of-school learning.’ The scope of out-of-school learning environments is quite broad, and these environments are often considered as ‘learning laboratories.’ The aim of this study is to determine the relationship between science teachers' self-efficacy beliefs and anxiety levels regarding out-of-school learning and the levels of science experience acquired by the secondary school students taught by these teachers in informal settings. A correlational research design was utilized as a quantitative method in the study. The study sample consisted of 100 science teachers working in state schools in Eskişehir during the academic year of 2022-2023 and 2,767 secondary school students attending these teachers' classes. Convenience sampling was adopted to determine the study sample. The following data collection tools were used: Teacher Self-Efficacy Beliefs Scale for Out-of-School Learning Activities, Anxiety Level Assessment Scale for Out-of-School Learning Environments, a personal information form, and the Science Experience in Informal Settings Scale. The findings indicate a significant relationship between science teachers’ self-efficacy beliefs regarding out-of-school learning and their ability to design and organize out-of-school learning activities. However, no significant differences in self-efficacy beliefs were observed with respect to training related to out-of-school learning environments, years of professional experience, or graduation background. Furthermore, science teachers’ levels of anxiety toward out-of-school learning did not differ significantly across any of the examined variables. A moderate negative correlation was identified between teachers’ anxiety levels and their self-efficacy beliefs. Finally, science teachers’ self-efficacy beliefs and anxiety levels were found not to significantly predict students’ science experiences acquired in informal learning settings.
In recent years, artificial intelligence has attracted great attention worldwide, particularly due to its impact on all areas of human life. The reasons for this interest include the rapid increase in the changes that artificial intelligence applications may bring to individuals’ lifestyles and the structure of society, and the potential problems that may accompany these changes. As in many other fields, artificial intelligence has quickly entered educational settings and has become a focal point for both students and teachers. In this respect, revealing teachers’ cognitive structures regarding artificial intelligence is of great importance, as their approaches to this development will directly influence their students. The aim of this study is to examine the conceptual frameworks of biology teachers regarding artificial intelligence. The data were collected from 126 participants using a free word association test. The analysis of the free word association test demonstrates various aspects of the participants’ conceptual frameworks concerning artificial intelligence. The participants emphasized the potential benefits of artificial intelligence in terms of advanced technologies, education, benefits to humanity, and contributions to the cultural and economic life of society. However, they also expressed the view that artificial intelligence may entail potential drawbacks, including ethical issues, security risks, and the risk of promoting laziness.
Artificial intelligence (AI) in science education offers significant advantages to pre-service teachers in lesson planning, developing teaching resources, and implementing personalized learning. Simulations, visualizations, and intelligent assessment systems contribute to the concretization of abstract concepts and increased student participation. This phenomenological study aims to determine preservice science teachers' views on AI. To this end, semi-structured interviews were conducted with 13 preservice teachers from different grade levels who had prior experience with AI, including virtual laboratory applications. A semi-structured interview form was used for this purpose, and the data were evaluated using content analysis. Strategies such as in-depth data collection, expert review, and member check were employed to ensure internal validity. The findings indicate that preservice teachers view AI as a tool that supports lesson planning and teaching processes. However, they also believe that excessive reliance on AI may limit teachers' creativity. Participants' views were grouped into the following categories: AI use in educational processes, the effects of AI on students, reliability, and the role of the teacher. One emerging finding is that preservice teachers often do not verify the information they obtain from AI. The study highlights the potential of AI as a pedagogical tool in science education and the factors teachers should consider when using AI.
This study examined the impact of the 'Systematic Planning Model', an ICT integration model, on student achievement and perceptions in a 4th-grade primary school science course. The research aimed to determine whether there were significant differences in the achievement and retention scores between an experimental group, where a curriculum based on this model was implemented, and a control group, where it was not. Additionally, students’ views on the model were explored. Using an explanatory mixed-method design, the study involved 45 students from a private school in Konya during the 2022–2023 academic year. Data were collected through an achievement test (reliability: 0.896) focused on the unit “Lighting and Sound Technologies from Past to Present” and semi-structured interviews. The findings revealed statistically significant improvements in the achievement and retention scores of the experimental group, while no significant changes were found in the control group’s scores. Qualitative results showed that students responded positively to the model, particularly appreciating the use of technology and interactive activities. They described the lessons as enjoyable and educational. Furthermore, many students suggested that similar technology-supported activities be implemented in other subjects, underlining the effectiveness of engaging, tech-integrated learning environments.
Today, knowledge and technology are produced rapidly. In this case, the aim of educators is to develop appropriate environments instead of presenting information directly to students and to develop students' ability to access and use information by blending it with other information. This situation reveals the importance of concept education in education with technological tools. In this study, it is aimed to understand what and how technological tools have been used to identify or eliminate misconceptions in physics education in the last 10 years. For this reason, 83 studies, including 22 theses and 63 articles, were analyzed in this field between 2010 and 2020. The studies were examined according to the themes of publication year, publication type, purpose, method/pattern, sample, data collection tools, technological tool method/technique used, data analysis method, and the subject studied. As a result of the examinations; it was seen that the studies in this field increased as the year 2020 approached, that they were aimed at eliminating rather than detecting, that the majority of the studies consisted of articles, that they focused on the abstract concepts of physics, which are difficult, that animations and simulations were mostly used, and that quantitative research methods were preferred.
The poles are recognized as the northernmost and southernmost points on Earth. These extremes are defined differently in terms of cartography, magnetism, geography, and the polar star. In Turkey, numerous activities have been carried out concerning the polar regions, such as the establishment of a scientific research camp and meteorological station in Antarctica, seabed mapping, and similar initiatives. These activities have continued to expand rapidly. In this context, it is anticipated that the targeted educational content will be more easily accessed through studies conducted not only by public and private research and development institutions but also within the field of education. This study aims to increase middle school students' awareness of 20 animal species living in the polar regions, as part of the “World of Living Things” unit and the “Let’s Get to Know Living Things” section of the Science curriculum. The research was designed using a qualitative approach, specifically the phenomenological design, which aims to understand the essence of participants lived experiences with a particular phenomenon. The participants consisted of 105 volunteer 5th-grade students from a public middle school in Ankara. Data were collected through open-ended, self-assessment forms and analyzed using qualitative methods. As part of the study, 20 animals living in the polar regions were identified, and an Artificial Intelligence-Based Educational Game (AIEG) was developed. The game was created using the image recognition feature on the Teachable Machine (URL1) platform and deployed as an online, web-based educational tool. After engaging with the game, students completed a self-assessment form. Additionally, qualitative data were obtained through question–answer interviews with students, and the findings were evaluated using content analysis. The results revealed that students initially had limited knowledge of animals living in the polar regions, but their awareness significantly improved after interacting with the educational game. Based on the findings, it is recommended that educational games be more widely integrated into school settings to support effective and lasting learning. Furthermore, this research is expected to serve as a model for future studies aimed at raising awareness about polar animals and contributing to the protection of species threatened by environmental challenges.
This systematic literature review examines the impact of generative artificial intelligence (AI) applications on student learning outcomes in middle school science education. Twelve studies that met the inclusion and exclusion criteria were included in the study. The studies were accessed from the WOS, SCOPUS, ERIC, and TrDizin indexes and databases. The PRISMA protocol was used in the selection of studies. The analysis reveals that generative AI tools significantly contribute to academic achievement, conceptual understanding, scientific and digital literacy, personalized learning opportunities, and students' motivational and emotional outcomes. The systematic review further emphasizes the importance of integrating AI tools into science education in a pedagogically and ethically sound manner. Recommendations include increasing AI literacy among teachers, incorporating ethical awareness into instruction, and developing culturally sensitive and interdisciplinary educational practices. Generative AI stands out as a transformative technology for improving the quality, accessibility, and inclusivity of science education.
The importance of virtual reality in science education is constantly increasing. For this reason, this study aims to identify current trends in science education through virtual reality (VR) themes using a bibliometric analysis of the Web of Science (WoS) database. Data analysis was conducted with VOSviewer software. Findings reveal that VR studies are primarily categorized under “Education Educational Research” and “Education Scientific Disciplines.” VR technology first appeared in science education research in 2002, with publications steadily increasing over time. The most prolific and influential researchers were Lamb, Richard, and Etopio, Elisabeth, while the United States, Australia, and Türkiye had the highest academic impact. The most frequently used keywords were “virtual reality,” “science education,” “augmented reality,” and “higher education,” while recurring abstract terms included “technology,” “knowledge,” and “research.” These results underscore the rising importance of VR technologies in science education and map the evolving research landscape, offering valuable insights for future studies and educational practices.
This study is developing a Deep Learning model automating the coding of drawings students provide about climate change phenomena in our world, as a learning contribution through formative assessment. We started first with ResNet50 architecture, but ultimately, we settled on MobileNetV2 reduced architecture for the sake of being able to integrate with mobile- and web-based applications. The challenge is the model has very few examples in the training set to work with, so we decided augmenting the data (i.e., rotate, zoom, flip horizontally,) will help the model generalize more reliably. The model achieved training accuracy of 92% and validation accuracy of 90%. Moreover, we were able to reduce the model size about 85% through optimization. Our model outputs not a simple classification, it also produces explanatory feedback for each class, and we have made possible for the feedback to be read by the student about their idea. Our findings are indicating, it is possible to use AI-based systems to teach how to investigate integrated fields like environmental education. Future studies will include multi-label classification, explainable AI (XAI) methodologies and dataset sizes will also increase.
