抄録
With advancements in generative artificial intelligence (AI), including large language models (LLMs), AI technologies are being increasingly utilized in various aspects of daily life, and their adoption is expected to proliferate. However, only a few studies have evaluated the content created by generative AI and human responses based on real-world scenarios. Assessing the quality of content involving generative AI and its impact on human psychological acceptance is crucial for appropriate utilization of these technologies. In this study, we evaluated the content created by generative AI for book endorsements. Specifically, the quality of the content created was compared under four generation processes: (i) solely by humans, (ii) solely by ChatGPT, (iii) by ChatGPT and then revised by humans, and (iv) by humans and then revised by ChatGPT. In addition, we investigated how disclosing the content generation process affects evaluation. The results revealed that AI content has a more significant effect on increasing purchase intention than content generated solely by humans. However, when the content-generation process was explained, some resistance toward AI-generated content was observed. This resistance was mitigated when humans were involved in certain aspects of the generation process.
