Consciousness, Mind Perceptions, and Emotional Reactions to AI: A Mixed-Methods Study Across Five Countries

AI systems are becoming more advanced and increasingly filling in human-like social roles with potential consequences for trust, ethics, and governance. Yet, little is known about how these beliefs are structured beyond exclusively Western samples. In a preregistered cross-national study (Australia, India, South Africa, the UK, and the USA, total N = 2,470), we examined the relationships among mind perceptions (agency and experience), current and future consciousness attributions, and emotionality of descriptions of AI. The results showed that perceiving AI as having agency and experience predicted consciousness attributions: experience perceptions were the dominant predictor of current consciousness, whereas agency was a slightly stronger predictor of future consciousness attributions. While samples from the five studied countries differed in how much they currently ascribe consciousness to AI, these differences narrow in future ascriptions, with respondents across all countries converging toward the belief that AI will be conscious in five years. Furthermore, higher levels of mind perception correlate with more emotionally charged descriptions of AI. The emotionality of individual descriptions correlates weakly with that of societal descriptions, and significant differences across countries (particularly pronounced in India and South Africa) are observed. Overall, the results suggest that the belief in an AI mind may be a key driver of its social classification, highlighting the relevance of these beliefs for how societies may approach AI governance.