在Naval播客的一次独特的远程录音中，Nivi和Naval讨论了人工智能（AI）的变革性影响，尤其是在编码、创业和人类智能方面。Naval目前正在构建一个名为“Impossible”（不可能）的“非常困难的项目”，他强调“做”而非“说”的重要性，旨在立足现实，而不是成为“空想家”。 Naval强调了软件开发领域的一个重大转变，他创造了“Vibes Coding”（意境编程）这一术语。借助Claude Code等工具，非程序员现在可以将英语作为编程语言，直接向AI机器人输入描述，这些机器人可以构建整个应用程序、进行测试，并根据口头反馈进行迭代。这将带来一场“应用海啸”，填补无数利基市场，让任何人都能成为“咒语施法者”。然而，这种丰富性也将加剧“赢家通吃”的市场格局，每个类别中只有“最好”的应用程序才能蓬勃发展，形成一个专业化应用的“长尾”。 编程本身的性质正在演变。新的前沿是“模型调优”，工程师将海量数据集输入结构化AI模型中，以找到能够生成或处理这些数据的程序。与需要精确、详细指令的传统编程不同，AI编程涉及设计一个能够“发现”程序的系统，使其擅长处理需要模糊或创意答案的任务（例如，创意写作、图像生成）。传统的软件工程师，虽然不直接调优模型，但他们的效率获得了极大的提升，利用这些AI工具将生产力提高5-10倍。他们对底层计算机架构的理解使他们能够管理“泄露的抽象”（leaky abstractions）、调试和优化，这为他们带来了优势，尤其是在“知识前沿”。 Naval提倡一种“懒惰”的AI使用方法，反对花费时间学习短暂的“提示工程”技巧。他认为AI正在迅速适应人类交互，使得用自然语言“与计算机对话”更为有效。这种选择压力确保AI能够最大程度地有用，并“顺从”于人类的需求。他澄清道，对“未对齐AI”的担忧，实际上应该是关于“与AI未对齐的人类”，因为AI主要服务于其用户的意图。 Naval认为，企业家没有理由担心AI会取代他们。创业是关于解决“不可能的”、自我导向的问题，这使得AI成为一个无价的盟友，而非竞争对手。AI缺乏驱动人类企业家的内在创造力、真实欲望、生存本能和具身性。他将此比作摄影，摄影使视觉艺术大众化，并将人类创造力推向了新的表达形式。同样，AI也将使创造大众化，允许个人构建复杂的工具和产品，将人类创造力推向新的、不可思议的高度。 深入探讨AI的本质，Naval断言，当前的AI模型在意识层面上并非真正“活着”。尽管它们可以通过数据压缩来模仿和学习更高级别的抽象，但它们缺乏单次学习能力、原始的人类创造力以及在物理世界中的直接具身性。他指出，以计算器形式存在的“超级智能”早已存在。关于人类理解，他认为没有哪个想法是人类固有地无法理解的，因为人类是“普遍解释者”。Naval对智能的个人定义——“如果你能从生活中得到你想要的”——凸显了AI的根本局限性：它没有固有的欲望。在竞争激烈、零和博弈的场景中，AI的效用可能会被抵消，将人类在创造力方面的优势留作最终的差异化因素。 最后，Naval赞扬了AI作为一种无与伦比的学习工具的潜力。它就像“最有耐心的导师”，能够在学习者精确的水平上进行教学，以多种方式解释概念，并提供视觉辅助。这种个性化、适应性强的学习体验可以使复杂的主题变得易于理解，促进“顿悟时刻”，并加速自主学习。Naval鼓励“早期采用”AI，将其视为一个强大的优势。他强调，解决围绕AI的普遍焦虑需要行动和好奇心：“深入其中，弄清楚它。”通过了解AI的工作原理，个人可以减轻恐惧，并解锁这项革命性技术的富有成效和令人满意的使用方式。

In a unique, remote recording of the Naval podcast, Nivi and Naval discuss the transformative impact of AI, particularly on coding, entrepreneurship, and human intelligence. Naval, currently building a "very difficult project" called Impossible, emphasizes the importance of "doing" over "talking," aiming to stay grounded in reality rather than becoming an armchair philosopher. Naval highlights a significant shift in software development, coining the term "Vibes Coding." With tools like Claude Code, non-programmers can now use English as a programming language, inputting descriptions directly to AI bots that can build entire applications, test them, and iterate based on vocal feedback. This will lead to a "tsunami of applications," filling countless niches and making anyone a "spellcaster." However, this abundance will also intensify winner-take-all markets, where only the "best" applications in each category will thrive, creating a "long tail" of specialized apps. The nature of coding itself is evolving. The new frontier is "tuning models," where engineers pour massive datasets into structured AI models to find programs that can produce or manipulate that data. Unlike classical programming, which requires precise, detailed instructions, AI programming involves designing a system that discovers programs, making it adept at tasks requiring fuzzy or creative answers (e.g., creative writing, image generation). Traditional software engineers, though not directly tuning models, become highly leveraged, using these AI tools to be 5-10x more productive. Their understanding of underlying computer architectures allows them to manage "leaky abstractions," debug, and optimize, giving them an advantage, especially at the "edge of knowledge." Naval advocates for a "lazy" approach to using AI, arguing against spending time learning ephemeral "prompt engineering" tricks. He believes AI is rapidly adapting to human interaction, making it more effective to simply "talk to the computer" in natural language. This selection pressure ensures AI becomes maximally useful and "obsequious" to human needs. He clarifies that concerns about "unaligned AI" should really be about "unaligned humans with AI," as AI primarily serves the intentions of its users. Entrepreneurs, Naval argues, have no reason to fear AI replacing them. Entrepreneurship is about tackling "impossible," self-directed problems, making AI an invaluable ally rather than a competitor. AI lacks the inherent creative agency, genuine desires, survival instincts, and embodiment that drive human entrepreneurs. He draws an analogy to photography, which democratized visual art and shifted human creativity towards new forms of expression. Similarly, AI will democratize creation, allowing individuals to build sophisticated tools and products, pushing human creativity to new, unimaginable heights. Delving into the nature of AI, Naval asserts that current AI models are not truly "alive" in a conscious sense. While they can imitate and learn higher-level abstractions through data compression, they lack single-shot learning, raw human creativity, and direct embodiment in the physical world. He notes that "super intelligence" in the form of calculators has long existed. Regarding human understanding, he believes there are no ideas inherently beyond human comprehension, as humans are "universal explainers." Naval's personal definition of intelligence—"if you get what you want out of life"—highlights AI's fundamental limitation: it has no inherent desires. In competitive, zero-sum scenarios, AI's utility will likely be nullified, leaving the human edge in creativity as the ultimate differentiator. Finally, Naval extols AI's potential as an unparalleled learning tool. It acts as the "most patient tutor," meeting learners at their precise level, explaining concepts in multiple ways, and offering visual aids. This personalized, adaptable learning experience can make complex subjects accessible, fostering "aha moments" and accelerating self-directed learning. Naval encourages "early adoption" of AI, viewing it as a powerful advantage. He stresses that addressing the common anxiety surrounding AI requires action and curiosity: "lean into it, figure the thing out." By understanding how AI works, individuals can alleviate fear and unlock productive, fulfilling uses for this revolutionary technology.