Decision-Making Preferences in AI Era: Human Perception of AI in Text-Based Tasks Including Emotional Context

Significant advancements have been made in the field of artificial intelligence (AI) in various domains, ranging from decision making to creative tasks. This study embarks on a comprehensive exploration of human-AI interaction, trust in AI-generated content and the capabilities of AI in emotional tasks. Applying an economic experiment, the paper assesses the preferences between human- and AI-generated responses in email communication with emotional context. Subjects made decisions under conditions of rating and comparing the emails. The scores given to the emails provided a metric for capturing the subjects’ preferences. Three hypotheses are explored: first, that humans surpass AI in understanding complex emotions; second, that knowledge of AI origins leads to evaluator bias against it, even when AI exceeds human performance; and third, that AI’s content, especially in emotional contexts, is rated lower due to doubts about its authenticity. The results confirmed the first two hypotheses, highlighting humans’ superior generation of emotional content and an evident bias against AI when its origins are known. Our findings open new avenues for future research, emphasising the multifaceted relationship between AI, human cognition, and emotion in text-based tasks, and shows that while AI’s potential in text generation is vast, its limitations become evident in the intricate terrain of human emotion.