Vercel's AI SDK offers a text-based version of its documentation specifically for LLMs, which is a thoughtful feature. Other documentation should consider adopting this approach.

This tool has been eye-opening. Could templates combined with an AI Code Agent be the future of website building?

This article provides an overview of AI agents, defining them as entities that perceive their environment and act upon it using tools. It emphasizes that the capabilities of an agent are determined by its environment, the actions it can perform, and the tools it has access to. The article discusses various types of tools, including those for knowledge augmentation, capability extension, and write actions. It also delves into the planning process, highlighting the importance of decoupling planning from execution and the role of reflection and error correction. The article explores different planning techniques, including function calling and hierarchical planning. It also discusses the challenges of tool selection and the importance of evaluating agents for planning failures, tool failures, and efficiency. The article concludes by emphasizing the significance of agents and their potential impact, while also acknowledging the need for further research and development in this field.

Chip Huyen's article on AI Agent is exceptionally well-written. It explains the concept clearly and accessibly, and also provides detailed, relevant examples. It's a bit lengthy, but trust me, it's worth the read.

An agent is like an assistant. You give it a task, such as washing the car, and it figures out the most effective way to accomplish it.

Here's how it might work:

• Plan: The agent researches the nearest car washes, their prices, and services.
• Action: The agent drives to the chosen car, ensures the car is washed, and then drives it home.
• Result: The task is completed.

In an AI agent, AI is the brain that processes the task, plans a sequence of actions to achieve this task, and determines whether the task has been accomplished.

Compared to non-agent use cases, agents typically require more powerful models for two reasons:

• Compound mistakes: an agent often needs to perform multiple steps to accomplish a task, and the overall accuracy decreases as the number of steps increases. If the model’s accuracy is 95% per step, over 10 steps, the accuracy will drop to 60%, and over 100 steps, the accuracy will be only 0.6%.
• Higher stakes: with access to tools, agents are capable of performing more impactful tasks, but any failure could have more severe consequences.

Complex tasks require planning. The output of the planning process is a plan, which is a roadmap outlining the steps needed to accomplish a task. Effective planning typically requires the model to understand the task, consider different options to achieve this task, and choose the most promising one.

This article is a personal reflection by a tenured professor, Matt Might, on his journey to tenure, prompted by a question about minimizing the risk of having a disabled child. Might shares his experience of having a son with a rare genetic disorder, which profoundly impacted his life and career. He argues that focusing on problems that matter, inspired by love and guided by knowledge, is more important than chasing metrics for tenure. He emphasizes the importance of teaching, service, and health, and regrets not pushing for open access earlier. He also highlights the crucial role of his wife in their family's journey and his success. Ultimately, he advocates for a holistic approach to academia, where making a difference is prioritized over traditional metrics.

This reminds me of river mussels that grow pearls around grains of sand. We've all experienced unfortunate events, perhaps even asking God, "Why me?" But instead of just accepting it, there's another option: Build your life around it. You're not the only one who has suffered; maybe you can do something to help others who have gone through something similar.

The author explores how to measure personal growth, moving beyond traditional metrics like net worth. Inspired by business growth metrics, she proposes three heuristics: rate of change (becoming a "new person" every 3-6 years), time to solve problems (tackling major life issues like career, family, and finance efficiently), and number of future options (making choices that maximize future possibilities). She uses the analogy of an investment, aiming to "double" herself by evolving and gaining new perspectives. The author emphasizes solving big problems to create safety nets for future exploration and prioritizing actions that expand future options. She acknowledges that these heuristics are tailored to her preference for novelty and exploration and may change in the future.

How can we tell if someone has grown over several years? Chip Huyen offers three tips. I particularly resonate with Point 2, "Time to solve problems," and Point 3, "Number of future options."

Big problems are like a foundation; they should be solid and reliable. As time goes on, they should become more robust, freeing you up to tackle more interesting challenges.

The number of future options is also an important measure. What opportunities are available to me today that weren't available five years ago?

In the face of uncertainty, I lean towards the decision that would give me the most future options. For example, I’d choose a job that pays less but gives me more job options in the future (e.g. if the job gives me exposure like allowing me to work on open source or publish papers). I’d prioritize tasks that teach me transferable skills instead of tasks that teach me niche, narrow skills.

While our approaches may vary, having clear metrics is crucial. These metrics will give us greater confidence in our purpose and help us persevere to achieve our goals.

This article discusses the core principle of Stoicism, which is to focus on what you can control and not worry about what you can't. It draws from the teachings of Epictetus and Marcus Aurelius, emphasizing the importance of dividing concerns into two categories: your actions and choices (the small bin you are responsible for) and everything else (the large bin that is the responsibility of the gods). The author explains that much of our suffering comes from trying to control the uncontrollable, and that true freedom and peace come from focusing on our own will and actions. The article also touches on the similarities between Stoicism and Buddhism, both of which emphasize letting go of the need to control everything. The author encourages readers to practice this principle daily, shrinking their sphere of concern to match their sphere of influence.

The core idea of this article, rooted in Stoicism, is to "focus your energy on what you can control and accept what you can't." While this sounds simple and practical, life is often more complex. For example, your job might bring pressure and unhappiness, even though it provides income. While you could control this situation, it would require significant effort. Ultimately, it comes down to identifying what matters most to you and then directing your energy accordingly.

This article introduces the concept that most things we buy has two prices: the first price, usually monetary, to acquire the item, and the second price, which is the effort and initiative required to actually use and benefit from it. The author argues that modern society focuses heavily on the first price, leading to a build-up of unused possessions and a feeling of unfulfillment. He suggests that we often overindulge in low-effort entertainment to avoid the second price, further depleting our time and energy. The solution is to consciously avoid needless first prices and focus on paying the second price for things we already own, which can lead to a more rewarding and fulfilling life. The author also mentions his "Depth Year" concept as a way to focus on using what we have.

I largely agree with the idea that most purchases have two prices. A prime example is "Read it Later" apps. We encounter an interesting article, click the button to save it, and instead of reading it, we immediately search for the next interesting article.

This reminds me of Goodhart's Law: "When a measure becomes a target, it ceases to be a good measure."

The conflict arises because the second "price" is paid by someone other than the initial purchaser. When you buy an ice cream, you don't just put it in the fridge and forget about it.

The initial purchase decision is usually driven by System 1, which is more emotional. If the actual consumption requires System 2, and no resources are allocated to it, then it will be postponed, perhaps indefinitely.

However, the author's proposed solution is too simplistic and impractical. It's like saying we all know that eating too much junk food is bad for us, so the solution is easy: just eat less junk food. It's far more complex. The entire industry is designed to appeal to our System 1, and once System 1 is making most of our decisions, it's difficult to resist its pull.

So I asked Gemini 2.0 Flash its opinion on more workable and complementary solutions. I find them more helpful.

• Start Small and Be Patient: Instead of trying to overhaul your entire life overnight, start with small, manageable changes. For example, commit to not buying any new books for a month, or dedicate 15 minutes each day to a hobby you've been neglecting. Gradual changes are more sustainable.
• Identify Your Triggers: Become aware of the situations or emotions that lead to impulse purchases. Once you know your triggers, you can develop strategies to avoid or manage them.
• Implement a "Waiting Period": Before making a non-essential purchase, wait a set period (e.g., 24 hours, a week). This can help you determine if you truly need the item or if it's just an impulse.
• Create a "Use It" List: Make a list of the things you already own that you want to use more. This can help you prioritize your time and energy.
• Schedule Time for Second Prices: Just like you schedule work or appointments, schedule time for activities that require effort and initiative, such as reading, learning a new skill, or exercising.
• Focus on Experiences: Shift your focus from acquiring material possessions to investing in experiences, such as travel, concerts, or spending time with loved ones. These experiences often provide more lasting satisfaction.
• Embrace Minimalism: Explore the principles of minimalism, which emphasize living with less and focusing on what truly matters. This can help you reduce clutter and prioritize your time and energy.
• Seek Community Support: Connect with others who are also trying to reduce their consumption and focus on second prices. Sharing your experiences and challenges can provide motivation and support.
• Reframe "Second Price" as "Investment": Instead of viewing the effort required to use something as a "price," reframe it as an "investment" in yourself and your well-being. This can make the process feel more rewarding.
• Practice Gratitude: Cultivate a sense of gratitude for the things you already have. This can help you appreciate what you own and reduce the desire to acquire more.
• Address Underlying Issues: If overconsumption is a symptom of deeper emotional issues, consider seeking professional help to address those issues.
• Focus on Progress, Not Perfection: Don't get discouraged by setbacks. The goal is to make progress, not to achieve perfection. Celebrate small victories and learn from your mistakes.

I tried the tricks mentioned in the article, and IT WORKS! It's like pure magic for me!

本文作者分享了过去一年使用大型语言模型（LLMs）进行编程的经验，并认为LLMs对他的工作效率有积极影响。他主要通过三种方式使用LLMs：代码自动补全、搜索复杂环境问题的答案以及聊天驱动编程。其中，聊天驱动编程虽然最具价值但也最复杂，需要调整编程习惯。作者强调，LLMs在处理明确目标和提供背景信息的任务时表现最佳，并建议将LLMs用于编写可轻松验证的代码，例如测试。此外，作者还探讨了LLMs如何影响代码结构，认为更小、更多的包可以更好地利用LLMs的优势。最后，作者介绍了正在开发的工具 sketch.dev，旨在为Go语言编程提供一个专门的LLM开发环境。

这也是我目前觉得使用 LLM 比较合适的姿势。基于 Code Base 去 generate code 虽然很方便，但容易失控。前期做好分解工作，给 AI 明确的、小的 Task，或者生成 Test Cases 这样繁杂的琐事可能更合适。

让 AI 总结了这段访谈，涉及到的内容还蛮多的。对于追溯童年印记、应对 AI 挑战、珍惜当下这几点颇有共鸣。

---

探索人生目标、权力动态与人际关系：与罗伯特·格林的深度对话

在一次引人入胜的访谈中，神经生物学家安德鲁·休伯曼与畅销书作家罗伯特·格林深入探讨了人生目标、权力动态、人际关系以及人类行为的复杂性。格林以其对历史、心理学和文化交织的独特见解而闻名，他分享了关于如何发现个人目标、驾驭人际关系以及理解人类动机的深刻见解。

发现人生目标：追溯童年印记

格林强调，人生目标并非一蹴而就，而是一个需要耐心和自我探索的过程。他认为，每个人在童年时期都拥有独特的“冲动之声”，这些声音引导我们走向特定的兴趣和天赋。他引用了心理学家霍华德·加德纳的“五种心智框架”理论，指出人类的智能并非仅限于智力，还包括语言、抽象、动觉和社会智能等多种形式。格林认为，我们的大脑会自然地倾向于其中一种或两种智能，而我们应该顺应这种倾向，因为这才是我们力量的源泉。

格林鼓励人们回溯童年，回忆那些令自己着迷的事物，并以此为线索，找到自己的人生方向。他分享了自己的经历，从小就对文字着迷，而这种原始的兴趣最终引导他走上了写作之路。他还引用了爱因斯坦、乔布斯和伍兹等名人的例子，说明他们从小就表现出对特定领域的强烈兴趣。

然而，格林也指出，随着年龄的增长，我们往往会受到来自老师、父母和同龄人的影响，逐渐迷失自我，不再倾听内心的声音。他强调，找到人生目标的关键在于重新与自己建立联系，倾听内心的声音，并理解自己的情绪，特别是挫败感和焦虑感，因为这些情绪往往在指引我们远离错误的方向。

权力动态：一种普遍的人类需求

格林将权力定义为一种普遍的人类需求，即对自身环境和人际关系拥有一定程度的控制感。他认为，这种需求并非总是负面的，而是源于我们对失控的恐惧。他指出，权力动态存在于各种人际关系中，包括师生、情侣和朋友之间。

格林强调，权力并非总是通过直接的方式表达，而是常常以微妙的方式进行，例如操纵、伪装和被动攻击。他认为，理解这些权力动态对于在社会和职业生涯中取得成功至关重要。他指出，他的著作《权力的48条法则》并非教人如何操纵他人，而是教人如何保护自己免受他人的操纵，并避免犯下常见的错误。

诱惑的艺术：一种复杂的人际互动

格林将诱惑定义为一种复杂的人际互动，其中一方通过吸引力、神秘感和情感上的脆弱性来影响另一方。他认为，诱惑的根源在于禁忌和违抗的欲望，以及我们对被他人接纳和理解的渴望。

格林指出，诱惑并非总是负面的，它也可以是一种积极的体验，例如当我们被一个故事、一本书或一个人的魅力所吸引时。他强调，被诱惑需要一定的脆弱性，而这种脆弱性本身也是一种积极的品质，因为它使我们能够与他人建立更深层次的联系。

浪漫关系：超越表面

格林认为，成功的浪漫关系需要建立在共同的价值观、兴趣和情感连接之上，而不仅仅是外貌或物质财富。他强调，真正的连接需要深入到对方的性格，了解他们的价值观、兴趣和生活方式。

格林还指出，幽默感在一段关系中至关重要，因为它可以帮助我们应对生活中的挑战，并保持彼此之间的吸引力。他认为，一段关系中应该存在一定的神秘感，这样才能保持新鲜感和吸引力。

非语言交流：一种被忽视的语言

格林强调，非语言交流在人际关系中至关重要。他认为，我们的大脑天生就对非语言信号非常敏感，但我们往往会忽略这些信号，而过于关注语言。他鼓励人们练习观察他人的肢体语言、面部表情和语调，因为这些往往比语言更能揭示一个人的真实想法和感受。

格林还指出，真正的微笑会点亮整个面部，而不仅仅是嘴巴。他认为，识别虚假微笑的能力对于避免与有毒的人建立关系至关重要。

人工智能的挑战：保护人类的独特性

格林对人工智能的快速发展表示担忧。他认为，人工智能可能会削弱人类的思考能力、自我意识和创造力。他强调，人类大脑是宇宙中最复杂的创造物，我们应该珍惜和发展它，而不是依赖人工智能来解决所有问题。

格林认为，我们应该将焦虑视为一种信号，它表明我们需要更深入地思考问题，而不是立即寻求简单的答案。他鼓励人们拥抱焦虑，并将其转化为创造力和行动力。

从逆境中成长：格林的个人经历

格林分享了他个人经历，包括他所经历的中风。他认为，中风虽然夺走了他的一些能力，但也让他对生活有了新的认识。他开始更加珍惜生命中的美好事物，并更加专注于自己的目标。

格林强调，我们应该珍惜当下，不要把生活视为理所当然。他鼓励人们拥抱脆弱性，并与他人建立更深层次的联系。

这篇文章探讨了互联网内容分发模式的演变，指出视频正在成为新的 HTML。文章认为，随着分发平台和内容模式的激增，内容呈现方式从传统的 HTML 和 JavaScript 转向更丰富的格式，如视频和即时应用。这种转变不仅改变了用户获取内容的方式，也影响了广告和数据分析。文章还强调了平台所有者对内容和指标的控制，以及移动广告拦截可能带来的挑战。最终，文章认为我们正在寻找一种新的移动运行时环境，而视频和富媒体内容正扮演着关键角色。

这篇文章写于 2016 年，确实有一定的前瞻性，一定程度上也确实在往这个方向发展。一图胜千言，一视频胜千图。视频可以承载的内容非常丰富，而且方便传播，用户接受度也高。但我觉得视频会让人变懒，刺激多巴胺的分泌，从 live a good life 这个角度来看，对视频的依赖长远来说一定是弊大于利。

这篇文章探讨了在现代社会中，人们越来越感到孤独，并提出了一种解决方案：搬到朋友附近居住。文章通过多个案例，包括 Toby Rush 和他的朋友们在堪萨斯城建立的紧密社区，以及 Phil Levin 创立的 Live Near Friends 平台，展示了这种生活方式的好处。文章强调了在朋友的支持下，生活可以变得更轻松、更有意义，尤其是在育儿方面。文章还提到了“孤独流行病”和“友谊衰退”等社会问题，并指出，与朋友保持紧密联系可以有效缓解这些问题。文章最后鼓励读者考虑这种生活方式，并强调其带来的便利和支持。

深有同感，现实世界能够经常联系的朋友，除了工作上的同事，往往就是住的比较近的朋友，这些经常联系的朋友形成的友谊网络，能够很好的克服孤独感和精神空虚。

这篇文章探讨了“魔法”背后的原理，并非超自然力量，而是投入了超出常人预期的努力和时间。文章以魔术师 Teller 的一个经典魔术为例，说明了为了达到看似不可能的效果，需要进行大量的提前准备和长时间的等待。这种原理不仅适用于魔术，也适用于其他领域，如音乐、演讲和艺术。文章强调，要实现看似不可能的目标，需要从小处着手，逐步积累，最终达到“一夜成名”的效果，而这背后是长期的投入和积累。

魔法的本质并非超自然，而是投入超出常人预期的努力和时间。

本文作者分享了其构建大型技术项目的方法，核心在于通过不断看到实际成果来保持动力。作者强调将大型项目分解为可快速产生可见结果的小任务，并优先构建可测试的组件。文章建议先构建“足够好”的子组件以快速推进到演示阶段，避免追求完美而阻碍进度。作者还提倡为自己构建软件，并尽快采用，以便根据实际使用情况迭代改进。文章总结了五个步骤，强调分解问题、快速演示、优先满足自身需求，并不断迭代。作者认为，这种方法适用于各种项目，并强调个人需要找到适合自己的激励方式。

对于自己的 Project，最重要的就是快速看到结果，然后获取反馈并不断优化。其中很重要的是分解能力：将大的项目分解成小的、可独立完成、可以测试并看到结果的子项目。

文中提到的一些技巧其实并不复杂，但个人项目容易夭折的一点就是很容易变成热情驱动，一旦热情下降就容易停滞，而一旦停滞就会一直停滞···

这篇文章探讨了通过迭代提示大型语言模型（LLM）“写更好的代码”是否能有效提升代码质量和性能。作者使用 Claude 3.5 Sonnet 模型，通过两种方式进行实验：一是随意地迭代提示“写更好的代码”，二是使用更明确的系统提示和用户提示，并加入“罚款”机制。实验结果表明，随意迭代虽然能逐步提升代码性能，但容易引入不必要的复杂性。而通过更明确的提示工程，可以更快地获得性能更优的代码，但同时也更容易出现细微的错误。文章还指出，LLM 在代码优化方面有其局限性，需要人工干预来识别和修复错误，并强调了 LLM 在提供新思路和工具方面的价值。最终，通过迭代提示和适当的提示工程，代码性能可以提升高达 100 倍。

简单来说，LLM 更喜欢明确的指令，抽象的指令（如写更好的代码）虽然可能 work，但可能会引入依赖或增加复杂度。如果把它比作车的话，目前还是半自动驾驶阶段，有时它会进入死胡同，此时就需要有经验的司机来指导它如何绕过去。因此要发挥这辆车的最大性能，司机的经验依旧不可或缺。

Of course, these LLMs won’t replace software engineers anytime soon, because it requires a strong engineering background to recognize what is actually a good idea, along with other constraints that are domain specific. Even with the amount of code available on the internet, LLMs can’t discern between average code and good, highly-performant code without guidance.

当然，这些大型语言模型在不久的将来不会取代软件工程师，因为这需要强大的工程背景来识别什么是真正的好主意，以及其他特定领域的约束。即使互联网上有大量的代码，LLM在没有指导的情况下也无法区分普通代码和优秀、高性能的代码。

本文探讨了互联网生态从早期的博客、论坛等分散模式向少数大型平台集中的转变，指出这种转变导致用户被锁定在“围墙花园”中，平台为了追求增长和盈利，往往牺牲用户利益。文章强调，技术熟练的用户应该考虑拥有自己的互联网“房产”，而非仅仅租用大型平台的空间。作者并非建议完全放弃大型平台，而是鼓励用户在利用这些平台的同时，建立自己的网站，并采取多元化的内容发布策略，以确保内容的可移植性和用户控制权。文章还提供了具体的建议，如拥有自己的域名和网站、避免过度依赖单一平台、以及在开放论坛中与受众互动等。

非常赞同作者的观点，这也是为什么博客在互联网巨头的垄断下依旧有存在的价值。巨头中的「异类」Telegram 也是个不错的选择。

这篇文章讨论了德国的“常客桌”（Stammtisch）传统，这是一种在酒吧或餐厅定期聚会，喝酒聊天的形式，被视为对抗孤独和促进人际关系的一种方式。Paul Graham 在 What I Worked On 这篇文章中也有提到过类似的定期小范围聚会形式：

One night in October 2003 there was a big party at my house. It was a clever idea of my friend Maria Daniels, who was one of the thursday diners. Three separate hosts would all invite their friends to one party. So for every guest, two thirds of the other guests would be people they didn't know but would probably like.

我觉得这种聚会非常棒，能够拉近人与人之间的关系，分享信息，有归宿感。很想组织一个这样的定期聚会，在杭州。

关于 SQLite 的一些冷知识。

• Open Source 但不 Open Contribute，需要被邀请，才能贡献代码。
• 每一行 SQLite 代码，差不多有 600 行对应的测试代码。
• 没有 Code of Conduct，但有 Code of Ethics，很有意思。
• SQLite 对于小文件的读写甚至比文件系统（fread, fwrite）还快（35% faster）。
• 有些情况下比 Redis 还快。
• 极致的兼容性。3.0+ 版本的 SQLite 都可以读写 3.0.0 版本的 SQLite（2004-06-18）
• SQLite 的作者没有找到一个合适的版本管理工具，就自己写了一个，名叫 Fossil，当然也是 SQLite 作为存储。

最近重温《康熙来了》的片段，依然觉得十分有趣。康永和小S的组合确实经典。然后就被推送了这个蔡康永的访谈视频。

过去只觉得康永说话让人感到舒适，但没有深究过这种舒适感从何而来。看完访谈后，感觉「尊重」在其中扮演着很重要的角色：他对嘉宾、观众、读者以及职业的尊重，都体现在他细致入微的言行中。他会设身处地地考虑他人的处境、情绪和状态，关注细节，并力求用最恰当的语言避免歧义和冒犯。比如，他会强调「看节目的人」而非「观众」，因为前者代表着一个个鲜活的个体，而后者则是一个模糊的群体。将个体视为群体，如同将活生生的人简化为数字，显然是不妥的。当然，这些是他对自己的要求，并非呼吁他人效仿。

特斯拉的这个 mini cybertruck 有点萌啊，1:43 比例，拧一拧发条就能跑，这谁能抗拒，当然也肯定是买不到的···

这种直播模式好像挺有意思的，国内的线下商家也可以借鉴下。

「胖东来」一直有耳闻，借着盗月社的视频简单感受了下，确实跟普通的超市不一样，感觉是盒马升级版。评论也很有料：

你应该拍一拍员工的😄，我去年去过一次，给我印象最深的其实是员工，他们状态真的不一样。工资，福利给的足，人家乐呵呵的对待顾客，而且不像其他超市都闲着假装拖地😂。

好像普通员工月收入七八千，每年有一个月带薪假，最近还加了几天的情绪假，就是心情不好也可以请假。

于老板无意间吐露了生活的真相：开心是因为你做了开心的事，而不是告诉自己很开心，做开心的事就会想做好，随之带来的是被别人称之为“责任”、“自律”或是“专注”等等的东西。我们追求或是求之不得的，或许都可以通过做让自己开心满意的事，让其自然而来。

胖东来的货不是最便宜的，甚至有些东西还比较贵，但是确实买得很放心。记得有一年有个哥哥结婚，婚房里的床上四件套是胖东来买的，他们直接派两个售货员到新房子去免费铺床，甚至在床上用大枣、花生、桂圆和莲子摆好了心形的“早生贵子”。

每个胖东来门口都有宠物寄存处，寄存处的猫狗笼子还有遮光帘，旁边的还有宠物饮用水以及一次性的宠物喂食碗。胖东来的电玩城抓娃娃，是可以让人帮忙调整位置，太久抓不上来是可以要求送一个的。

胖东来每周二闭店放假，而且是在员工正常休假基础上的，所以胖东来的员工一年有140+天假。 胖东来会标示商品“进价”，没错，是进价。 胖东来利润绝大部分用于分红给每个员工。

在 Github Copilot 的帮助下，很方便地写了一个 Music Player，满意。（数据源来自自己多年积累下来的曲库）

Finally someone made it!

Between 2009 and 2012, iPhones had a built-in "Send to YouTube" button in the Photos app. Many of these uploads kept their default IMG_XXXX filenames, creating a time capsule of raw, unedited moments from random lives.

Inspired by Ben Wallace, I made a bot that crawled YouTube and found 5 million of these videos! Watch them below, ordered randomly.

太喜欢这种「由 CEO 来演示产品使用」的 Marketing Strategy 了（不知道是不是受 Jason Fried 的影响），没有各种炫技的动画、Scroll Effect、光影效果，直接明了。

Eagle App 居然出自台北 90 后独立开发者之手，太强了。

孙正义在 2017 年的一个演讲，谈到了它对未来 30 年的判断，按照他的估计，超级人工智能（远超人类智能）会在未来 30 年内成为现实，因为人类的神经元数量、连接几乎不会发生太大的变化，而芯片上的晶体管数量会不断地增加（直到摩尔定律失效），再加上特定的软件、算法，计算机的智能超过人类只是时间问题，它认为 30 年内就会到达这个「奇点」。他的估算模型虽然不太精确，但从结果上看，在 2024 年的这个时间点，最 Top 的 AI 模型的智能毋庸置疑已经超过了人类智能的均值。

更进一步，孙正义认为随着芯片技术的不断迭代更新，它可以在很小的体积上集成高度的智能，这个智能体会被嵌入到生活的方方面面，从穿的鞋子，到开的汽车，到家用电器，到城市管理。这是一个巨大的市场，这也是软银收购 ARM 背后的一个考量。

孙正义的一些思考还是挺超前的，虽然看起来有点飘，但世界大致上也确实是在往那个方向发展。

Hacker News 的这份 Guidelines 太棒了，一些 tips 很值得社区经营者借鉴。比如：

• On-Topic: Anything that good hackers would find interesting. Off-Topic: Most stories about politics, or crime, or sports, or celebrities, unless they're evidence of some interesting new phenomenon.

• don't do things to make titles stand out, like using uppercase or exclamation points, or saying how great an article is.

• if the title contains a gratuitous number or number + adjective, we'd appreciate it if you'd crop it. E.g. translate "10 Ways To Do X" to "How To Do X,"

• Comments should get more thoughtful and substantive, not less, as a topic gets more divisive.

• When disagreeing, please reply to the argument instead of calling names. "That is idiotic; 1 + 1 is 2, not 3" can be shortened to "1 + 1 is 2, not 3."

• Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something.

• 主题相关：任何优秀黑客可能会感兴趣的内容。非主题相关：大多数关于政治、犯罪、体育或名人的故事，除非它们是某些有趣新现象的证据。

• 不要做让标题显眼的事情，比如使用大写字母或感叹号，或者夸赞文章有多好。

• 如果标题包含无意义的数字或数字+形容词，我们希望您能将其裁剪。例如，将“10 Ways To Do X”翻译为“How To Do X”。

• 随着话题越来越具有争议性，评论应该更加深思熟虑和有实质内容，而不是更少。

• 当不同意时，请回复论点而不是叫名字。“那太愚蠢了；1 + 1 等于 2，不等于 3”可以简化为“1 + 1 等于 2，不等于 3。”

• 请不要发布肤浅的否定，尤其是针对他人工作的。好的批评性评论能让我们学到一些东西。

再推一下沉浸式翻译的 Epub Bilingual Maker，制作双语 Epub 太方便了。

内容量很大的一期访谈，嘉宾 Dr. K 非常有料，他的 YouTube 上有很多不错的内容，可以一并看下。

本期访谈探讨的是比较 Fundamental 的话题：孤独感、上瘾、目标感的缺失这些现代人普遍会遇到的问题。信息量非常大，我完整地听了一遍，将音频 Transcribe 成文本后又读了好几遍。以下是压缩之后，我觉得挺值得记录的一些点：

• 多关注自己，创造与自己独处的时间与空间，多与自己沟通，更多地了解自己。不要被多巴胺诱导，过多地关注外部世界，玩身份游戏。
• 把每一次负面情绪、积极情绪当作了解自己的机会：为什么我会难过，为什么我会做这件事，驱使我做一件事的动力是什么。
• 多巴胺会带来短期快感；血清素会带给你幸福感、满足感和宁静感；催产素可以增进亲密关系、信任和同理心。多巴胺与血清素的数量成反比。
• 沉迷电子设备会带来孤独感、疏离感、影响两性关系和与自己的连接。
• 情绪共鸣（比如同时感到惊喜、害怕等）可以拉近人与人之间的联系。
• 自信源于内在（基于对自己的了解），自负来自外在（基于外界的反馈）。发自内心的信念会影响大脑，进而影响行为，别人能分辨内在的自信和外在的自负。如果我们失去与自己的连接，就失去了自信的基础。
• 情绪跟身体的关系密切，情绪会反映到身体，身体也可以影响情绪。
• 无法获得持久的幸福很重要的一点是，与外部同频而不是与内部同频。