Contextual Acceptance of Artificial Intelligence in Commercial Policy-Making: An Integrated Model of Acceptance, Resistance, and Institutional Factors

Abstract

This study examines how senior decision-makers in Iranian commercial policy institutions evaluate and respond to artificial intelligence (AI), with particular attention to the interplay of adoption, resistance, and institutional considerations. It develops the notion of contextual acceptance to describe the importance of whether AI-supported decisions can be justified to relevant stakeholders. An exploratory sequential mixed-methods design was used. The qualitative phase involved 18 semi-structured interviews with experts in commercial policy, information technology, and academia. The authors report thematic saturation after the fourteenth interview and four additional interviews to consolidate the themes. In the quantitative phase, a 27-item five-point Likert questionnaire was completed by 248 senior managers and specialists selected through stratified random sampling. The reported analysis used partial least squares structural equation modeling (PLS-SEM) in SmartPLS 4. The qualitative analysis identified digital leadership, perceived usefulness, institutional trust, algorithmic opacity, organizational resistance, and perceived risk. In the structural model, digital leadership (β = 0.36, t = 6.3), perceived usefulness (β = 0.34, t = 5.9), and institutional trust (β = 0.28, t = 4.7) were positively associated with attitudes toward AI. Algorithmic opacity (β = −0.22, t = 3.8) and organizational resistance (β = −0.19, t = 3.4) were negatively associated with attitudes. The reported R² values were 0.64 for attitude and 0.69 for behavioral intention, with an overall GOF value of 0.61. All nine hypotheses were reported as supported, although the indirect-effect coefficient and test statistic for mediation were not provided. The findings are consistent with a context-sensitive account of AI acceptance in which perceived institutional accountability and the defensibility of algorithm-supported decisions complement assessments of technical usefulness. Potential implications include clearer accountability arrangements, explainable AI systems, and digital leadership development; causal and policy-effectiveness claims require further study.