Adoption of generative artificial intelligence in chemistry and chemical engineering education: A PRISMA-guided scoping review of pedagogy, assessment, and governance

Generative artificial intelligence (genAI) has entered chemistry and chemical engineering education with unusual speed and unusually high epistemic stakes. Because chemistry learning depends on verbal explanation, symbolic notation, quantitative reasoning, visual representation, and safety-constrained practice, while chemical engineering adds process-scale models, flowsheets, piping and instrumentation diagrams, control logic, scale-up assumptions, and professional safety decisions, the educational consequences of large language models are more complex than a generic question of efficiency. This PRISMA-guided scoping review synthesizes a bounded corpus of 25 chemistry and chemical engineering studies and interprets it alongside broader higher-education scholarship on assessment, trust, policy, and responsible AI use. Across the included literature, adoption is unfolding along two linked paths: unsanctioned student use that destabilizes conventional assessment, and deliberate pedagogical integration that asks students to test, critique, verify, and sometimes redesign machine output. The review also foregrounds less positive impacts, including over-reliance, hallucinated evidence, unequal AI literacy, erosion of assessment validity, and unsafe transfer of AI-generated advice into laboratory or process contexts. Five themes recur: assessment disruption, prompting and AI literacy, chemistry- and chemical-engineering-specific capability limits, laboratory and teacher-facing integration, and governance. The review argues that genAI in chemistry and chemical engineering education should not be framed as a simple tutoring technology. Instead, it should be understood as a socio-technical infrastructure whose educational value depends on disciplinary verification practices, representational competence, professional judgment, and institutional design.