When self-control is no longer a protective factor: the moderating effect of institutionalized identity on attitudes toward artificial intelligence.

AI technology is increasingly embedded in multiple critical domains of social functioning, reshaping human work patterns, decision-making processes, and the allocation of responsibility. The aviation sector is particularly illustrative: from autopilot systems and intelligent flight-path planning to predictive maintenance and smart air traffic management, AI is progressively entering the core of flight decision-making and safety assurance (; ). International civil aviation organizations and national regulators have identified AI as a key direction for the digital transformation of aviation (). However, advances in technical capability do not automatically translate into full acceptance at the operational level. Individuals and groups differ substantially in their attitudes toward AI (), and these differences affect not only technology adoption behaviors but also the safety and effectiveness of human–machine collaboration (). In high-risk industries, practitioners' attitudes toward AI are often embedded in deeper considerations of responsibility attribution, risk consequences, and the boundaries of professional judgment. Ignoring these psychological and socio-structural factors may impede the effective deployment of AI, and in certain contexts could even introduce new systemic risks (). Understanding the differences in AI attitudes across occupational groups and their psychological bases () has therefore become an important agenda item in AI social governance. Existing research on the psychological mechanisms underlying AI attitudes often contains an insufficiently examined assumption: that individual psychological traits function consistently across different occupational and responsibility contexts. Is this premise valid? Do psychological regularities discovered in general populations apply equally in high-responsibility occupational settings? AI attitudes refer to individuals‘ overall evaluative orientations toward AI technology and its applications, typically comprising three interrelated components: cognitive, affective, and behavioral-intentional (). The cognitive component involves beliefs about AI's capabilities, reliability, and potential consequences; the affective component reflects emotional experiences when confronted with AI, such as feelings of security, anxiety, or perceived threat; the behavioral-intentional component encompasses tendencies to use, rely on, or avoid AI systems. Early research largely situated AI attitudes within the Technology Acceptance Model (TAM) or the Unified Theory of Acceptance and Use of Technology (UTAUT) framework (; ), focusing on how technology-attribute variables such as perceived usefulness and perceived ease of use influence attitudes and usage intentions. As AI has evolved from a tool-support technology into a system with autonomous, decision-participating capabilities, the limitations of these traditional acceptance models have become apparent (). Compared with general information technology, AI is more likely to trigger psychological concerns around autonomy, control, and responsibility attribution (; ), and recent scholarship has converged on trust as the central psychological pivot of human–AI engagement, with appropriately calibrated trust depending on individuals' technology self-efficacy, perceived risk, and dispositions toward automation ().