From "Ion-Adsorption" to "Ion-Adsorbed" Rare Earths: Terminological Drift, AI Paraphrasing, and the Erosion of Geochemical Precision

Olivier Pourret , Andrew Hursthouse , Karen Johannesson, Haiyan Liu

The rapid adoption of artificial intelligence (AI) tools for writing and paraphrasing is transforming scientific communication. While such tools can improve linguistic fluency, their misuse, particularly when employed to mechanically rephrase text, may introduce subtle but consequential distortions in scientific terminology. In disciplines such as geochemistry, where terminology is closely tied to processes and mechanisms, even minor linguistic shifts can alter conceptual meaning. This opinion paper examines a growing instance of terminological drift in the rare earth element (REE) literature: the increasing substitution of the established term ion-adsorption rare earths with the grammatically similar but conceptually misleading expression ion-adsorbed rare earths. Ion-adsorption rare earth deposits constitute a distinct class of REE mineralization formed through the adsorption of rare earth cations onto clay mineral surfaces during intense chemical weathering. The defining feature of these deposits is therefore a process "ion adsorption" rather than a static condition of the elements themselves. Through bibliometric analysis of the Dimensions database and examination of recent publications, we document the rapid diffusion of the alternative phrasing in the scientific literature since the mid-2010s, with a marked acceleration after 2020. Notably, several articles display internal inconsistencies in which the correct and incorrect formulations coexist within the same manuscript, a pattern consistent with partial automated paraphrasing or unsupervised text rewriting. We argue that replacing a process-defining noun (adsorption) with a past-participial adjective (adsorbed) subtly shifts interpretation from mechanism to state, thereby weakening the conceptual framework that underpins genetic models, extraction strategies, and environmental assessments of these deposits. More broadly, this case illustrates how automated paraphrasing tools, when applied without adequate domain expertise, may inadvertently erode terminological precision in process-driven sciences. Maintaining scientific integrity in the age of AI therefore requires collective vigilance from authors, reviewers, and editors to ensure that technological assistance does not compromise conceptual clarity or disciplinary rigor.