<![CDATA[AbstractGenerative artificial intelligence tools pose a fundamental challenge to traditional technology acceptance models, which were originally designed to evaluate static technologies and largely overlook the relational and interactive dimensions of AI-powered translation tools. This challenge is particularly pronounced in the context of learning linguistically complex languages such as Arabic, where constructs such as trust in output accuracy, perceived risk associated with over-reliance, and perceived intelligence of the tool emerge as primary drivers of adoption. To address this theoretical and methodological gap, the present study aims to develop and establish the content validity of a novel measurement instrument that moves beyond the limitations of conventional technology acceptance models and foregrounds human–AI interaction. Given the inherently fuzzy nature of these emergent constructs, the Fuzzy Delphi Method was employed to elicit systematic consensus from a panel of ten experts in Arabic language pedagogy and educational technology on an initial pool of 60 items. The analysis, applying a threshold of d ≤ 0.20 and a minimum agreement level of 75%, resulted in the retention of 58 items. Beyond offering a validated measurement instrument, this study proposes an alternative and more contextually appropriate theoretical framework for assessing AI adoption in language education, thereby opening new avenues for investigating the complex dynamics between learners and intelligent systems.]]>
