The Four Artificial Intelligence Apprehension Scales: Apprehension Towards Personal AI, General AI, Institutional AI, and Large Language Models

Recent research has developed measures that conceptualise human factors in relation to AI in relatively broad terms, such as general attitudes and positive, negative or adaptive use. However, more nuanced instruments are still needed to capture specific psychological dimensions, such as apprehension, which is the focus of this paper. Moreover, many existing measures conceptualise AI as a single, undifferentiated object of evaluation, rather than distinguishing between different AI systems, contexts of use, or functional domains. The present study adapted, developed and validated four distinct but structurally parallel instruments measuring apprehensions toward General AI, Large Language Models, Personal AI and Institutional AI. The instruments share identical structure and item content adapted to each type of AI. Apprehension is operationalised across three dimensions (Implied Malice, Undesirability, and Unpredictability). Data from a British sample of 559 adults (age range 18–45, M = 30.64, SD = 6.75, 50.3% males). For each scale, confirmatory factor analyses supported the three-dimensional structure, while internal consistency was strong at both total and subscale levels, and model-based indicators demonstrated well-defined latent constructs. The scales demonstrated good discriminant and convergent validity. These findings establish four distinct and psychometrically robust instruments suitable for research requiring measurement of apprehension toward artificial intelligence both at a general level and across specific classes of AI systems.