AI-assisted Simulations for Inquiry-Based Chemistry Education Evaluation

Thesis Abstract

The rapid expansion of digital learning environments has elevated the role of AI-driven simulations in chemistry education, yet systematic evaluation of their effectiveness in supporting inquiry-based learning remains limited. This study addresses the gap by examining how AI-assisted simulations influence learners’ inquiry competencies, conceptual understanding, and motivation within high-school and undergraduate chemistry contexts. The aim is to evaluate not only learning outcomes but also the pedagogical processes that AI simulations enable, including hypothesis generation, experimental design, data analysis, and evidence-based reasoning. Specific objectives are to (1) assess gains in inquiry-based skills using AI-simulated experimentation, (2) determine the impact on content mastery across foundational chemistry topics, (3) examine student engagement and motivation as mediated by perceived realism and interactivity, (4) investigate teacher facilitation practices that optimize AI-simulation use, and (5) develop a practical framework for integrating AI-assisted simulations into inquiry-based curricula. A quasi-experimental mixed-methods design will be employed across two mid-sized tertiary institutions, involving 240 participants 120 in the experimental group using AI-assisted simulations for guided inquiry and 120 in the control group employing traditional laboratory activities and conventional simulations. The study will adopt socio-constructivist and situated learning perspectives, drawing on Kolb’s experiential learning theory and Vygotsky’s zone of proximal development to frame analyses of learner interaction with AI-driven environments. Data collection will include pre- and post-tests assessing domain knowledge (e.g., chemical thermodynamics, kinetics, and stoichiometry), validated instruments measuring inquiry skills (e.g., Hypothesis Formulation, Experimental Design, Evidence Evaluation scales), and motivation metrics (e.g., Science Motivation Questionnaire II). Instrument validity will be established through expert panel review and pilot testing (n=40), with reliability confirmed via Cronbach’s alpha (? ? .80). Complementary data will be gathered from classroom observations (VIDEO-logged for coded indicators of inquiry discourse), think-aloud protocols (n=40 sessions), and semi-structured interviews with teachers (n=12) and a subset of students (n=40). Quantitative analyses will include ANCOVA to compare post-test gains between groups while controlling for pre-test scores, multivariate ANOVA to examine domain-specific outcomes, and regression analyses to identify predictors of inquiry skill development and motivation. Mediation analysis will explore whether perceived authenticity and interactivity mediate the relationship between AI-simulation exposure and inquiry outcomes. Qualitative data from think-aloud protocols and interviews will be analyzed using thematic analysis, following the Braun and Clarke steps, to extract patterns related to hypothesis generation, evidence-based reasoning, and instructional support needs. Integration of quantitative and qualitative results will utilize a convergent parallel design, with triangulation to validate inferences about the efficacy and mechanisms of AI-assisted simulations. Expected findings include statistically significant improvements in inquiry skill measures and content mastery for the experimental group, with effect sizes of moderate magnitude (Cohen’s d ? 0.40–0.60). It is anticipated that higher levels of perceived authenticity and interactive complexity will positively predict inquiry outcomes, while supportive teacher facilitation will strengthen the relationship between AI-simulation use and student engagement. The study is likely to reveal nuanced differences across student subgroups (prior achievement levels and learning preferences) and identify best-practice features for AI-simulations, such as guided hypothesis generation prompts, real-time data analytics, and scaffolded experiment design templates. The study contributes to knowledge by providing robust empirical evidence on the instructional value of AI-assisted simulations for inquiry-based chemistry education, articulating a theoretically grounded framework for their integration, and delineating programmatic guidelines for teachers and curriculum designers. It extends the application of Kolb’s and Vygotsky’s theories into AI-enabled inquiry environments and offers a validated instrument set for future evaluations. Recommendations emphasize scalable, context-responsive implementation, professional development for teachers in AI-facilitated inquiry modeling, and iterative refinement of simulation affordances to maximize student-driven investigation, data interpretation, and scientific argumentation. The study concludes that AI-assisted simulations, when embedded within well-structured inquiry pedagogy and supported by deliberate teacher facilitation, can meaningfully enhance both the process and product of chemistry learning, justifying broader adoption and ongoing research into adaptive AI features that personalize inquiry experiences.