The development of students’ abilities to analyse and interpret chemical processes is a central goal of contemporary science education, particularly in the context of competency-based curricula and an increasing emphasis on scientific reasoning. Despite systematic instruction, many lower secondary school students experience persistent difficulties in constructing evidence-based explanations, interpreting experimental data, and transferring chemical knowledge to unfamiliar or problem-oriented contexts. These challenges indicate a need for pedagogical approaches that explicitly support reasoning processes while also addressing learner diversity. In recent years, artificial intelligence (AI) technologies have attracted growing attention as tools capable of enhancing personalisation, feedback, and formative assessment in classroom practice. However, their effective integration requires a clear pedagogical framework and a well-defined role for the teacher.

This study explores the pedagogical potential of integrating the Claim–Evidence–Reasoning (CER) framework, problem-based learning (PBL), and AI-supported instructional tools to foster students’ analytical and interpretative skills in chemistry education at the lower secondary level. The CER framework was selected as it provides an explicit structure for scientific argumentation, enabling students to formulate claims, support them with appropriate chemical evidence, and justify the relationship between evidence and claims through scientific reasoning. Problem-based learning was employed to situate learning within meaningful and cognitively challenging contexts, encouraging students to engage in inquiry, hypothesis generation, and explanation of chemical phenomena. AI technologies were incorporated as supportive tools to enhance differentiation, scaffold reasoning processes, and provide timely formative feedback.

The theoretical foundation of the study draws on constructivist learning theory and inquiry-based science education, which emphasise active knowledge construction through engagement with problems and evidence. Within this framework, AI was conceptualised not as a substitute for teaching, but as an adaptive pedagogical assistant that supports the teacher’s instructional decisions. This position aligns with contemporary perspectives on educational innovation, including the view articulated by Sal Khan, founder of Khan Academy, who argues that artificial intelligence can function as a personalised tutor for every learner, enabling scalable individual support. At the same time, the study reflects the position of Andreas Schleicher (OECD), who highlights that while AI will not replace teachers, teachers who effectively use AI are likely to transform educational practice.

The intervention was implemented during regular chemistry lessons and focused on topics involving the interpretation of chemical reactions, experimental observations, and cause–effect relationships. Learning activities included problem scenarios, incomplete texts, data interpretation tasks, and CER-structured writing assignments. AI tools were used to support lesson planning, generate tiered tasks, and provide adaptive prompts and feedback templates for formative assessment. All instructional decisions, feedback validation, and summative assessments remained under the teacher’s control.

The findings indicate that the combined use of CER, PBL, and AI tools contributed to measurable improvements in students’ ability to analyse chemical processes, select relevant evidence, and construct coherent explanations. Additionally, increased student engagement and participation were observed during problem-solving activities. The results suggest that this integrated approach offers a promising pathway for developing higher-order thinking skills in chemistry education while maintaining the central role of the teacher.

The study was designed as a practice-oriented pedagogical intervention conducted in a lower secondary school chemistry classroom. The methodological framework was informed by principles of design-based research, allowing for the systematic implementation, observation, and refinement of instructional strategies in an authentic educational context. The participants were students enrolled in compulsory chemistry courses, representing a heterogeneous group in terms of prior achievement and learning needs. The intervention was implemented over a sequence of chemistry lessons focused on the analysis and interpretation of chemical processes, including reaction mechanisms, experimental outcomes, and qualitative observations. Instructional activities were designed according to the principles of problem-based learning. Each learning sequence began with a contextualised problem situation requiring students to explain a chemical phenomenon or interpret experimental data. Students worked individually and collaboratively to propose explanations and test their ideas using chemical knowledge. The CER framework was systematically embedded into learning tasks and classroom discourse. Students were explicitly taught how to formulate claims, identify and select appropriate chemical evidence, and articulate reasoning that links evidence to claims using scientific concepts. CER templates and guiding questions were provided, particularly at the initial stages of the intervention, and gradually reduced to promote learner autonomy. Artificial intelligence tools were used to support instructional design and formative assessment. These tools assisted the teacher in generating differentiated tasks, creating adaptive prompts for students at different levels of proficiency, and developing structured feedback aligned with the CER components. AI-generated outputs were critically reviewed and adapted by the teacher to ensure curricular alignment and scientific accuracy. Importantly, AI was not used for automated grading; instead, it functioned as a resource for enhancing feedback quality and instructional efficiency. Data collection methods included classroom observations, analysis of student written work, and formative assessment records based on CER criteria. Students’ responses were analysed to identify changes in the quality of claims, the relevance of evidence used, and the coherence of reasoning. The data were analysed qualitatively, with attention to patterns of improvement in analytical and interpretative skills. Ethical considerations were addressed by ensuring that participation was embedded within regular instructional practice and that student work was analysed anonymously for research purposes.

The results of this study demonstrate that the integrated use of the CER framework, problem-based learning, and artificial intelligence technologies can effectively support the development of students’ analytical and interpretative skills in chemistry education. By explicitly structuring scientific reasoning through the CER approach and situating learning within meaningful problem contexts, students were better able to articulate explanations of chemical processes and justify their conclusions using appropriate evidence. The findings further suggest that AI technologies can enhance instructional practice when used as supportive tools rather than as replacements for teacher expertise. In this study, AI contributed to improved differentiation, more efficient lesson preparation, and richer formative feedback, while the teacher retained full responsibility for pedagogical decisions and assessment. This balanced integration supports contemporary views on the role of AI in education, emphasising augmentation rather than substitution. From a pedagogical perspective, the study highlights the importance of aligning technological innovation with well-established learning theories and instructional frameworks. The effectiveness of AI tools was closely linked to their integration within the CER and PBL structures, underscoring the need for purposeful and theory-informed use of technology. The study has implications for chemistry teachers and curriculum developers seeking to promote higher-order thinking and scientific reasoning. It also contributes to the broader discourse on responsible and meaningful integration of artificial intelligence in science education. Future research may extend this work by examining long-term impacts on learning outcomes, exploring student perceptions of AI-supported learning, and investigating scalability across different educational contexts.

1. Berland, L. K., & Reiser, B. J. (2009). Making sense of argumentation and explanation. Science Education, 93(1), 26–55. 2. Hmelo-Silver, C. E. (2004). Problem-based learning: What and how do students learn? Educational Psychology Review, 16(3), 235–266. 3. Khan, S. (2023). Brave New Words: How AI Will Revolutionize Education (and Why That’s a Good Thing). Viking. 4. OECD. (2019). OECD Future of Education and Skills 2030. OECD Publishing. 5. OECD. (2021). Digital Education Outlook: Pushing the Frontiers with AI, Blockchain and Robots. OECD Publishing. 6. Schleicher, A. (2018). World Class: How to Build a 21st-Century School System. OECD Publishing. 7. Toulmin, S. (1958). The Uses of Argument. Cambridge University Press. Zohar, A., & Dori, Y. J. (2003). Higher order thinking skills and low-achieving students. Journal of the Learning Sciences, 12(2), 145–181.