AI-assisted laboratory notebooks to enhance inquiry-based science learning

Thesis Abstract

The rapid expansion of digital learning tools in science education has highlighted a gap between students’ experiential inquiry and the documentation of their laboratory reasoning. Conventional lab notebooks often fail to capture the iterative nature of inquiry, limiting opportunities for meta-cognition, feedback, and scalable assessment. This study investigates AI-assisted laboratory notebooks (AILN) as a technology-driven solution to enhance inquiry-based science learning by automating data capture, guiding experimental design, and providing real-time, evidence-based feedback to learners. The aim is to determine whether AILN improves students’ scientific reasoning, data literacy, and self-regulated learning compared with traditional pen-and-paper practices. Specific objectives are (1) to examine the impact of AILN on students’ ability to articulate hypotheses, document procedures, and interpret data; (2) to evaluate changes in inquiry self-efficacy and metacognitive awareness; (3) to assess teacher and student perceptions of usability, engagement, and instructional alignment with inquiry-based learning standards; (4) to identify contextual factors that influence successful adoption of AILN in secondary and tertiary science courses; and (5) to propose a scalable implementation framework for integrating AI-assisted notebooks into standard curricular workflows. A mixed-methods research design will be employed, combining quasi-experimental and case-study approaches. The quantitative strand will involve a pretest–posttest control group design with 240 participants drawn from three secondary schools and two undergraduate science courses across a metropolitan district. Participants will be stratified by grade level and prior achievement. The independent variable is the use of AILN versus traditional notebooks, while dependent variables include measures of scientific reasoning (using the Lawson Classroom Test of Scientific Reasoning), data literacy (a validated Data Literacy Scale), inquiry self-efficacy (Inquiry Self-Efficacy Scale), and metacognitive awareness (MAIA-3). Data will be analyzed using ANCOVA to control for baseline differences, followed by multilevel modeling to account for nested data structures (students within classes). Qualitative data will be collected from 30 semi-structured teacher interviews, 12 focus groups with students, and 10 classroom observations. Thematic analysis will be conducted to identify patterns related to user experience, instructional alignment, and perceived constraints. Instrument validity and reliability will be established through content validity indices, pilot testing, and Cronbach’s alpha coefficients above 0.80 for all scales. The AI-assisted notebooks will integrate natural language processing, automatic code-free data tagging, and rule-based guidance that prompts researchers to document hypotheses, procedures, controls, observations, and interpretations. The system will provide real-time formative feedback grounded in constructivist and DSQ (Diagnostic Sequencing of Questions) principles, while ensuring data provenance and ethical considerations through auditable AI logs. The theoretical framework combines Vygotsky’s social constructivism to situate collaborative inquiry within scaffolded AI-supported environments and Bandura’s self-efficacy theory to explain changes in students’ confidence in conducting scientific investigations. The study will also draw on the Technological Pedagogical Content Knowledge (TPACK) framework to examine teacher readiness and curricular fit. Anticipated findings include (i) statistically significant improvements in scientific reasoning, data literacy, and inquiry self-efficacy for the AILN group relative to controls; (ii) higher quality evidence sequences in students’ laboratory notes, demonstrated by richer hypothesis articulation, explicit linkage between methods and results, and causal inferences supported by data; (iii) positive user experiences with high perceived usability and perceived alignment with inquiry-based learning objectives, albeit with challenges related to initial setup, digital equity, and data privacy. The study will contribute to knowledge by providing empirical evidence on AI-enabled documentation systems as catalysts for deeper inquiry, offering a robust analytical model for evaluating learning gains in technologically augmented laboratories, and proposing a scalable framework for integrating AI-assisted notebooks into diverse science curricula. Conclusions are expected to indicate that AILN can enhance inquiry-based science learning by improving documentation quality, fostering metacognitive reflection, and strengthening data interpretation skills, while maintaining essential teacher facilitation. Recommendations include professional development for educators on AI-assisted inquiry scaffolding, strategies to mitigate equity gaps in access to technology, guidelines for data governance and student privacy, and a phased implementation roadmap aligned with national science standards.