以下是播客节目“This Week in Startups”的总结，重点关注关键讨论和主题： 节目开始，Jason Calacanis 和 Alex Wilhelm 回顾了节目的历史，然后迅速转向他们的第一位嘉宾，来自 Slow Ventures 的 Sam Lesson。 谈话的中心围绕着 Slow Ventures 新推出的 6000 万美元 Creator Fund。 Lesson 解释了该基金的理论基础：投资于首先建立社区，然后再在其上构建产品的创作者，有效地颠倒了传统的创业模式。 他强调了在这些社区中低客户获取成本 (CAC) 和高生命周期价值 (LTV) 的重要性。 Lesson 澄清说，该基金并非仅仅专注于创作者平台或个人产品投资，而是专注于与创作者的深度合作。 Calacanis 强调了 Mr. Beast 作为成功利用其受众的创作者的典范，并指出 Mr. Beast 的 "Feastables" 产生的收入可能甚至超过广告收入。 Lesson 告诫不要关注像 Mr. Beast 这样的一般娱乐创作者，他认为拥有高价值客户的利基社区更适合他们的投资策略。 他还详细介绍了该基金的投资结构：100 万至 300 万美元用于创作者现金利润的 10% 股权，在某种程度上类似于收入分成，但旨在协调激励机制并允许创作者拥有灵活性。 Calacanis 分享了他与 TikTok 创作者 Chef Reactions 的合作经验，他曾孵化过这位创作者。 他强调了理解利基垂直领域内特有的广告和产品的重要性。 讨论还涉及了潜在的挑战，例如对单一创作者的依赖（“玛莎·斯图尔特问题”）以及定义投资协议中包含的内容。 Lesson 强调了他们方法的整体性，投资于创作者的整体业务并信任他们的判断。 他们相信 "risen tis"，即魅力（riz）和对特定利基市场的专注痴迷（autism）的结合。 下一部分嘉宾是 Helion 首席执行官 David Kirtley，Helion 是一家聚变能源初创公司。 Kirtley 解释了 Helion 的技术，该技术旨在直接从聚变反应中提取电力，而不是使用热量来煮沸水并运行蒸汽涡轮机。 他概述了他们的目标，即到 2028 年产生第一批电子，并在 2030 年通过与微软的购电协议实现满功率。 该协议激励 Helion 达到这些目标，否则可能会成为一种负担。 他将聚变与核能进行了对比，强调了聚变的安全性优势，即可以通过停止燃料供应来关闭聚变反应。 最后一部分的嘉宾是编辑总监 Harison New 和 Lon Harris。 话题是人工智能的版权影响，这是由 Thomson Reuters 战胜 Ross Intelligence（一家较早的法律人工智能初创公司）的法律胜利引发的。 Jason 提出了四部分的合理使用测试。 他们探讨了对 OpenAI 和 Stable Diffusion 等生成式人工智能公司的潜在影响，这些公司使用大量的受版权保护的数据来训练他们的模型。 Lon 认为这些模型本质上是反版权的，呼吁人工智能公司与内容创作者达成许可协议。 New 强调了潜在的经济影响，表明该法律判决可能会削弱人工智能公司的估值。 Calacanis 和 Harris 争论了这些公司是否能够负担得起许可费用。 Calacanis 和 Lon 预测，未来人工智能可以引用来源来表明模型经过重新设计并注明信息来源。

Here's a summary of the podcast episode "This Week in Startups," focusing on the key discussions and topics: The episode began with Jason Calacanis and Alex Wilhelm discussing the show's history and then quickly moved to their first guest, Sam Lesson from Slow Ventures. The conversation centered around Slow Ventures' new $60 million Creator Fund. Lesson explained the fund's thesis: investing in creators who build communities first and then layer products on top of them, effectively inverting the traditional startup model. He emphasized the importance of low customer acquisition cost (CAC) and high lifetime value (LTV) in these communities. Lesson clarified that the fund wasn't solely focused on creator platforms or individual product investments but on deeply aligned partnerships with creators. Calacanis highlighted Mr. Beast as a prime example of a creator successfully leveraging their audience, noting that Mr. Beast's "Feastables" might generate even more revenue than advertising. Lesson cautioned against focusing on general entertainment creators like Mr. Beast, arguing that niche communities with highly valuable customers were a better fit for their investment strategy. He also detailed the fund's investment structure: $1-3 million for 10% equity of the creator's cash profits, resembling revenue sharing to some extent, but designed to align incentives and allow creators flexibility. Calacanis shared his experience with a TikTok creator, Chef Reactions, whom he incubated. He emphasized the importance of understanding endemic advertising and products within niche verticals. The discussion also addressed potential challenges, such as dependency on a single creator ("the Martha Stewart issue") and defining what's included in investment deals. Lesson stressed the holistic nature of their approach, investing in the creator's overall business and trusting their judgment. They believe in "risen tis," a combination of charisma (riz) and focused obsession (autism) for the particular niche. The next segment featured David Kirtley, CEO of Helion, a fusion energy startup. Kirtley explained Helion's technology, which aims to directly extract electricity from fusion reactions instead of using heat to boil water and run a steam turbine. He outlined their goal of generating first electrons by 2028 and achieving full power by 2030 through a power purchase agreement with Microsoft. This agreement has incentives for Helion to meet those targets, or it could be a liability. He contrasted fusion with nuclear, highlighting the safety advantages of fusion, which can be turned off by stopping the fuel flow. The final segment featured Harison New, editorial director, and Lon Harris. The topic was the copyright implications for AI, sparked by a legal victory by Thomson Reuters against Ross Intelligence, an older legal AI startup. Jason brought up the four-part fair use test. They explored the potential ramifications for generative AI companies like OpenAI and Stable Diffusion, which train their models on vast amounts of copyrighted data. Lon argued that these models are inherently anti-copyright, calling for licensing agreements between AI companies and content creators. New highlighted the potential economic implications, suggesting that the legal decision could chip away at the valuations of AI companies. Calacanis and Harris debated whether these companies could afford to pay for licenses. Calacanis and Lon predicted a future where AI could cite sources to show the models were reengineered and attribute the information.