The landscape of interactive systems is shifting toward dynamic, generative experiences that empower users to explore and construct knowledge in real time. Yet, timelines -- a fundamental tool for representing historical and conceptual development -- remain largely static, limiting user agency and curiosity. We introduce the concept of a generative timeline: an AI-powered timeline that adapts to users' evolving questions by expanding or contracting in response to input. We instantiate this concept through KnowledgeTrail, a system that enables users to co-construct timelines of historical events and knowledge formation processes. Two user studies showed that KnowledgeTrail fosters curiosity-driven exploration, serendipitous discovery, and the ability to trace complex relationships between ideas and events, while citation features supported verification yet revealed fragile trust shaped by perceptions of source credibility. We contribute a vision for generative timelines as a new class of exploratory interface, along with design insights for balancing serendipity and credibility.

翻译：交互式系统的发展正转向动态生成式体验，使用户能够实时探索和建构知识。然而，时间线——作为表征历史与概念发展的基础工具——在很大程度上仍保持静态，限制了用户的自主性与探索欲。我们提出了生成式时间线的概念：一种由人工智能驱动的、能够响应用户输入进行扩展或收缩，从而适应用户动态问题的时间线。我们通过KnowledgeTrail系统实现了这一概念，该系统使用户能够协同构建历史事件与知识形成过程的时间线。两项用户研究表明，KnowledgeTrail能够促进好奇心驱动的探索、激发偶然性发现，并帮助用户追踪思想与事件间的复杂关联，同时其引用功能虽支持信息验证，但也揭示了由信源可信度认知塑造的脆弱信任关系。我们提出了将生成式时间线作为新型探索性界面的愿景，并提供了平衡偶然发现与信息可信度的设计洞见。