Purpose Model Simulation - Purpose Formation of Multi-stakeholder by Dialog with LLM-Based AI

The increasing complexity of social and industrial structures emphasizes the importance of collaborative projects with diverse stakeholders in areas such as urban development and digital service development. Implementing AI-based digital services requires considering the impact on a wide range of stakeholders, including developers, users, AI data providers, and workers affected by the rise of AI. While incorporating diverse perspectives can enhance service value, aligning the goals and participation of all stakeholders presents challenges, including conflicts of interests and expectations. The "Purpose Model" framework was developed to visualize the roles and objectives of various stakeholders, fostering mutual understanding and aligning common goals and direction. However, identifying all stakeholders and ensuring their fair participation is difficult, with some struggling to clearly express participation and objectives. The rapid expansion of Large Language Model (LLM)-based conversational AI services, such as ChatGPT, offers the potential to simulate perspectives of less active stakeholders by mimicking specific human personas. Although this approach can provide valuable insights, the information generated by AI may not always be accurate or unbiased. This paper presents a comparative study using LLM-based AI to simulate the review process of the Purpose Model in multi-stakeholder co-creation projects, testing the feasibility of comprehensive stakeholder identification and role analysis. The study acknowledges the need for a combination of human and AI-driven reviews to ensure inclusivity and comprehensiveness in stakeholder engagement in co-creation projects.