“It Can’t Be That Simple”: exploring Finnish non-experts’ conceptions of AI and whether these predict emotional orientation towards AI

Abstract As artificial intelligence (AI) becomes increasingly embedded in everyday life and rises on the political agenda, research on how the public thinks and feels about it becomes increasingly relevant. Drawing on cognitive appraisal theory, which holds that emotions arise from people’s interpretations of external phenomena, this study investigates whether Finnish non-experts’ conceptions of AI relate to their emotional orientations. Prior to a public AI lecture series, 141 Finnish adults completed a survey. Open responses to “What is AI?” and “How AI works?” were abductively coded into eight features spanning technical, capability, and relational conceptions. Emotional orientations were measured with four Likert items (excitement, fear, personal relevance, trust) and analyzed using hierarchical clustering. Logistic regression tested whether conception features predicted profile membership. Most common conception features were Autonomous (72%), Data-driven (66%), Anthropomorphic (53%), Human-controlled (40%), and Rule-based (31%). Additionally, 37% expressed uncertainty. Four emotional profiles emerged: Optimists (22%), Cautious (28%), Disengaged (26%), and Indifferent (24%). Conceptions only weakly predicted orientations: anthropomorphic and probabilistic framings increased the likelihood of Optimist membership, but model accuracy was modest. The findings underscore the need for AI literacy initiatives that address both diverse misconceptions and emotional diversity. Limitations include convenience sampling in one country and challenges in coding anthropomorphism, which is partly embedded in technical terminology. Future work should further explore the role of anthropomorphic and probabilistic conceptions in shaping emotional orientations towards AI.