The paper "See Widely, Think Wisely: Toward Designing a Generative Multi-agent System to Burst Filter Bubbles" addresses the issue of "filter bubbles" formed by AI-powered search and recommendation systems, which reinforce users' biases and narrow their perspectives. The authors propose a human-centered approach to explore how Large Language Models (LLMs) can assist users in embracing diverse perspectives. They developed a prototype featuring LLM-powered multi-agent characters that users can interact with while reading social media content. The prototype includes features such as a frictionless interaction flow, a Viewpoints Jigsaw Puzzle, and progressive viewpoints sequence, along with gamification design to motivate user engagement. A participatory design study with 18 participants found that multi-agent dialogues with gamification incentives could motivate users to engage with opposing viewpoints. Progressive interactions with assessment tasks promoted thoughtful consideration. The study also identified three key design considerations: providing diverse perspectives through multi-agent characters, fostering deliberate and critical thinking through progressive interaction and assessment tasks, and motivating user engagement through natural interaction and gamification design. The prototype was evaluated through a user study, which revealed that participants generally found the system interesting and engaging. The multi-agent characters and gamification design were particularly well-received. However, some participants noted concerns about the credibility of the AI agents and the need for more detailed responses. The study also showed that the prototype facilitated deeper contemplation and exploration of diverse perspectives, supporting the effectiveness of the proposed design approach in helping users burst their filter bubbles.The paper "See Widely, Think Wisely: Toward Designing a Generative Multi-agent System to Burst Filter Bubbles" addresses the issue of "filter bubbles" formed by AI-powered search and recommendation systems, which reinforce users' biases and narrow their perspectives. The authors propose a human-centered approach to explore how Large Language Models (LLMs) can assist users in embracing diverse perspectives. They developed a prototype featuring LLM-powered multi-agent characters that users can interact with while reading social media content. The prototype includes features such as a frictionless interaction flow, a Viewpoints Jigsaw Puzzle, and progressive viewpoints sequence, along with gamification design to motivate user engagement. A participatory design study with 18 participants found that multi-agent dialogues with gamification incentives could motivate users to engage with opposing viewpoints. Progressive interactions with assessment tasks promoted thoughtful consideration. The study also identified three key design considerations: providing diverse perspectives through multi-agent characters, fostering deliberate and critical thinking through progressive interaction and assessment tasks, and motivating user engagement through natural interaction and gamification design. The prototype was evaluated through a user study, which revealed that participants generally found the system interesting and engaging. The multi-agent characters and gamification design were particularly well-received. However, some participants noted concerns about the credibility of the AI agents and the need for more detailed responses. The study also showed that the prototype facilitated deeper contemplation and exploration of diverse perspectives, supporting the effectiveness of the proposed design approach in helping users burst their filter bubbles.