The multidimensional structure of risk: how dread and controllability shape attitudes toward artificial intelligence

Artificial intelligence has spurred large-scale innovation, affecting politics, the economy, and society in unpredictable ways. How then do ordinary citizens perceive AI and its risks? We propose that perceived dread and controllability concerns are central to understanding public opinion about AI and its associated risks. This article introduces a theoretical framework synthesizing these dimensions and validates novel measures – the AI Dread and AI Controllability Concern Measures – using original surveys fielded in two distinct cases: Canada and Japan. Our findings reveal a multidimensional structure of AI risk attitudes, with key cross-national predictors of dread and controllability concerns including individuals’ trust in scientists, conspiracy thinking, and beliefs about technological change negatively affecting their job prospects. We encourage researchers to adopt these multi-item measures in their work on AI and its relationship to society, either as explanatory or outcome variables.