In conversational communication, misunderstandings often occur due to a lack of transmitted information. This problem would be important in online communication, such as the metaverse, where non-verbal information is limited. To address this issue, we developed a system for visualizing conversations in the metaverse by applying the functions of image generation AI. This system extracts keywords from the user's conversation, creates prompts for image generation based on the keywords, and then generates image from the prompts and displays it in the metaverse space. Experiment was conducted to verify the accuracy of the generated image using this system, and the results showed that the initial description did not generate an accurate image due to a lack of information, but as explanations were added, the system generated an image that matched the user's thought. Furthermore, evaluation experiment was conducted to validate the effect of consensus building between two users by using this system, and the results showed users understood the other's thoughts from the visualized images, revised their own thoughts, and reached consensus by using the proposed system effectively.