Generative artificial intelligence in university chemistry education: a critical review using Johnstone's chemistry triplet and Biggs' 3P model

Abstract

Generative artificial intelligence (GenAI) is entering university chemistry faster than its educational consequences can be evaluated. Chemistry makes this transition unusually demanding because learning depends on coordinated reasoning across observable phenomena, molecular and particulate models, symbolic notation, quantitative relations, experimental evidence, and safety-constrained practice. This critical review draws on structured searches of Scopus, ACS Publications, the Journal of Chemical Education, ERIC, Google Scholar through Publish or Perish, and Web of Science Core Collection, with a targeted update through 21 August 2026. It synthesizes representative empirical studies, performance benchmarks, classroom implementations, and contextual scholarship without claiming an exhaustive systematic-review corpus. Johnstone's chemistry triplet provides the primary disciplinary framework, while Biggs’ Presage-Process-Product model is used as a secondary pedagogical lens. The literature considered here indicates that GenAI can support critique, formative dialogue, scientific writing, representational comparison, coding, experimental design, and chemical engineering problem-solving when students retain executive control and verify outputs. The same systems can generate plausible but chemically incorrect mechanisms, unstable calculations, fabricated references, incomplete procedures, and unsafe advice. Evidence remains concentrated in short-term benchmarks, perception studies, and bounded classroom implementations; durable transfer, retention, representational fluency, and professional judgment are rarely tested. Responsible adoption therefore requires verification-centered assessment, explicit AI literacy, transparent reporting of models and prompts, human moderation, and chemistry-specific outcome measures. GenAI is most defensible as a fallible object of disciplinary inquiry, not as an autonomous source of chemical truth.