Using machine learning to investigate consumers' emotions: the spillover effect of AI defeating people on consumers' attitudes toward AI companies

Purpose The purpose of this study is to investigate consumers' emotional responses to artificial intelligence (AI) defeating people. Meanwhile, the authors investigate the negative spillover effect of AI defeating people on consumers' attitudes toward AI companies. The authors also try to alleviate this spillover effect. Design/methodology/approach Using four studies to test the hypotheses. In Study 1, the authors use the fine-tuned Bidirectional Encoder Representations from the Transformers algorithm to run a sentiment analysis to investigate how AI defeating people influences consumers' emotions. In Studies 2 to 4, the authors test the effect of AI defeating people on consumers' attitudes, the mediating effect of negative emotions and the moderating effect of different intentions. Findings The authors find that AI defeating people increases consumers' negative emotions. In terms of downstream consequences, AI defeating people induces a spillover effect on consumers' unfavorable attitudes toward AI companies. Emphasizing the intention of helping people can effectively mitigate this negative spillover effect. Practical implications The authors' findings remind governments, policymakers and AI companies to pay attention to the negative effect of AI defeating people and take reasonable steps to alleviate this negative effect. The authors help consumers rationally understand this phenomenon and correctly control and reduce unnecessary negative emotions in the AI era. Originality/value This paper is the first study to examine the adverse effects of AI defeating humans. The authors contribute to research on the dark side of AI, the outcomes of competition matches and the method to analyze emotions in user-generated content (UGC).