Understanding the Effects of AI Agents on Human
Aki Nagano · Academy of Management Proceedings · 2025
With the rise of generative artificial intelligence (AI), the application of AI agents has grown rapidly in numerous fields. Prior studies found that the effectiveness of AI in organizations depends on individuals' cognitive and emotional trust in the technology. Human-like AI is perceived as more trustworthy by users. Understanding user perceptions of AI interactions is critical for effective corporate strategies. This study aims to analyze the interaction between humans and generative AI through a comparative lens, based on the hypothesis that people tend to perceive generative AI more positively because human-like AI is seen as more trustworthy than traditional AI. Consequently, this study examines how individuals’ perceptions of AI evolve from traditional AI to generative AI. The method utilizes a before-and-after approach with a text-mining analysis, employing sentiment analysis and Term Frequency-Inverse Document Frequency (TF-IDF) techniques. The data is collected from subscribers' opinions in major Japanese newspapers. Sentiment analysis results showed that subscriber sentiment had statistically significant changes before and after the introduction of ChatGPT. This suggests that following the launch of ChatGPT, negative and positive values are rising, indicating an increase in the intensity of emotions. Some TF-IDF results show that it depended on the editor's choices and favored specific popular AI themes among subscribers. Their perceptions also tend to lean towards Japanese customs. Cultural factors must be considered when introducing the AI agent, as they influence perceptions of AI-human interaction.