Brushing my teeth is a good time to do squats. I did 30 this morning. #minihabit
Haoriwa retweeted
我不能允许中文圈的朋友错过这篇文章!读完大概8分钟,这可能是10月份最值得的8分钟。
我给大家翻译好了👇
如何变得更有雄心:写给与模型协作的人类使用指南
阅读时长:约 8 分钟
我在 Claude 团队工作。这是一段很有意思的经历:几乎每天,我都会发现有人借助 Claude 完成了某项壮举,而这些成就往往是我未曾考虑过、曾认为不可能做到,或者单纯因为缺乏好奇心和时间而未曾尝试的。就在这个月,它可能参与协助发现了一颗新行星;我看到了第一支自己真心喜欢的 AI 音乐视频;我们甚至还在讨论数学的新纪元。
模型的能力并非均衡发展。正如数学家陶哲轩(Terence Tao)所言,模型“在某些方向令人惊叹,而在另一些方向却显稚嫩”。有些人成功驾驭了这些惊人的能力,仿佛翱翔于天际——看着大家用我参与打造的产品攀登难以置信的高度,十分令人着迷。但还有许多人仍在努力适应他们的新羽翼,困惑于除了询问“法国的首都是哪里”或“帮我把这封邮件润色得别那么生气”之外,模型到底还能做什么。
我和其他从事 AI 工作的人经常呼吁:在使用 AI 时,我们应当更有雄心(ambitious)。这已经成为了一句集结号角式的口号。每当我们看到令人赞叹的应用时,就会转发并写道:“我们的野心还不够大。”但我深知将这种建议付诸实践有多么棘手,因此,这里有一份关于“如何变得更有雄心”的操作指南。
我认为“雄心”可以分为四种维度:速度(Speed)、高度(Altitude)、深度(Depth)以及广度(Range)。
速度(Speed): 用更少的时间完成同样的工作,这是 AI 经济价值的基本盘:如果你需要写代码,我用 Claude 能完成得更快。这一点已被广泛认知,且不需要在使用 AI 的方式上发挥太多创造力,因此本文略过不谈。
高度(Altitude): 提升抽象层级,关注最终结果而非单一具体任务。
深度(Depth): 像行业顶尖专家一样去审视和推演任务。不再停留于任务表层,而是深入一层,触及底层的原始数据、工具和根本问题。
广度(Range): 借鉴多个领域的经验与方法,而非仅仅局限于任务表面所需的那一个领域,去调用那些你平时通常不会用到的知识、工具和技术。
我主要使用 Claude(显而易见),但我的建议并不受限于特定模型。
高度(Altitude)
我们常随口将其称作“让模型在更高的抽象层级上运行”。当你给模型分配一项任务时,它执行的不过是你对“如何达成目标”所做的最优猜测。要在“高度”上更有雄心,不妨问问自己——接下来你希望发生什么?目标背后的终极目标是什么? 去年年初,我还会让 Claude 帮我排查并修复单个软件崩溃;而到了后来,我直接要求 Claude 确保我们的应用程序“零崩溃”。
让我们以一个日常(甚至可以说是琐碎)的办公任务为例。假设你在一家名为 Acme Inc 的虚构公司,需要为季度业务回顾(QBR)制作一套汇报幻灯片(PPT)。如果你输入一堆数据,Claude 很乐意为你生成几张幻灯片;如果你提供模板,它也可以顺畅填入内容。但这很难与“发现新行星”相提并论。来看看一些具体的步骤,让我们的思路逐层提升雄心:
基础步骤: 大多数业务回顾都想知道本季度发生了什么变化、原因是什么、以及应该采取什么对策。克制住仅在模板里更新数字的冲动——尽可能把你能获取到的所有原始数据喂给模型,让它理解你的业务逻辑:“告诉我本季度发生了什么变化,背后的原因是什么,以及我们应该如何应对。”
进阶步骤: 会议(至少高效的会议)最终都应产出明确的决策。向 Claude 提问:“基于你目前了解到的所有情况,梳理出眼前最紧迫的三个决策是什么,并针对潜在的发展方向展开分析,分别列出正反两面的论据。”
高阶步骤: 季度业务回顾诞生于这样一个时代:那时需要人类耗费大量精力筛选数据、厘清数据背后的含义,并判断公司是否需要做出业务调整。你可以这样要求 Claude:“业务评审的存在是为了让管理层及早发现问题,把资金和人力调度到正确的位置。请设计一种比‘季度汇报 PPT’更好的机制来实现这一点,并结合我们以往的历史数据证明该机制能更早发现问题;同时让这套新方案与当前机制并行运行一个季度。”
“高度”策略在追求某种终极成果时效果最好,但在纯创意类任务中表现稍逊。判断何时适合在“高度”上发力的一个显著标志是:你是否需要为了手头正在做的事情去开会讨论。 当然,目前这一方法仍有局限(在 2026 年,“Claude,帮我去赚钱”对绝大多数公司来说单凭一句话还是行不通的),但我普遍发现:随着模型能力的增强,它们处理高层级抽象任务的表现也越来越出色。
“高度”也是构建持久工作流(将人类移出执行闭环、无需你反复输入一项又一项任务)的关键所在。要做好这一点,通常需要重构你的工作流程——为了让 Claude 能够自主确保应用程序零崩溃,我首先得保证它能轻松拉取崩溃报告,理解哪个应用版本对应哪次 Git 提交,并赋予它访问 Windows、macOS 和 Linux 环境的权限,使其能够真正独立地复现并验证修复方案。
深度(Depth)
有时候,你只是想把某一项具体任务执行得极其出色。在“深度”上拥有雄心,意味着促使模型去使用该领域最顶尖专家所采用的工具、技术、媒介和专业知识——并持续深挖,直到连那些专家都能从你的产出中获得启发。
继续沿用 Acme Inc 的季度业务回顾例子。假设 Acme 公司正面临用户流失(随着时间推移客户流失)的困境。我发现结合以下技巧非常有效:
坚持使用业内顶尖专家所使用的工具: 当被要求“表现得像个专业人士”时,模型往往只是换了套行业术语,而没有真正改变工作方法。要在深度上解决问题,我们必须使用顶尖数据科学家和统计学家同款的工具。当 Claude 倾向于生成简单的 Markdown 表格时,坚持要求它构建数据库和统计模型。你不需要预先知道专家用什么,Claude 通常心里有数。
设定看似离谱的目标,并告诉模型这完全合情合理: 你需要将网页加载时间从 10 秒缩短到 5 秒?不妨直接要求达到 200 毫秒。如果模型觉得某件事不可能办到,重申这完全是可解的,它只需要继续探索尝试。
挑战既有认知: 尤其在商业领域,许多我们坚信不疑的事情其实只是对现实的粗糙近似。将原始信息与数据提供给模型,让它以对抗性的视角去检验既有假设:“我们一直认为用户流失是由于上手引导(onboarding)体验不佳导致的。请尝试从原始数据中复现并验证这一推论。”
创造新知识: 这一点可以顺理成章地承接对既有认知的挑战——一旦 Claude 已经深入数据腹地,让它去探寻盲区:“基于系统日志构建一套能够复现去年业务情况的客户模拟系统。然后找出我们尚未察觉的最早流失信号,并证明该信号能在未参与训练的新数据上完成有效预测。”
引入外部裁判: 切勿让模型对自己打分评判。我相信随着模型智能的提升,这一问题终会解决,但在 2026 年 10 月的当下,我强烈建议引入外部验证机制。最理想的方式是提供一个可供模型直接运行并实时获取反馈的评估工具。
我经常使用的一个简短提示词(Prompt):
“我希望这项任务更加深入,而不是范围更大。在重写任务前,请先列出:(1) 该领域的顶尖专家会做哪些我目前没有要求的事情;(2) 他们会从哪些最原始的材料入手,而不是使用已被加工整理过的数据;(3) 他们会使用哪些具体的方法和专业工具;(4) 除了你和我之外,还有什么客观指标或外部手段能证明结果是正确的。完成以上分析后,重写我的提示词,精确提出这些要求。”
广度(Range)
简而言之,在“广度”上拥有雄心,意味着我们从多个不同领域的视角来审视手头的工作——引入比过去在成本考量下所能承受的更多跨界学科。这并不是多么独到的洞见,每一部盗贼冒险电影(Heist movie)在前 30 分钟里都在忙着集结性格迥异、各怀绝技的团队成员。随着 AI 工具让多智能体系统(Multi-agent systems,包含多线程、编排器或单纯的子智能体)的运行变得越发轻而易举,要求 Claude 跨界引入额外领域往往能带来惊艳的效果。
将其付诸实践并不简单,需要你发挥创造力。直到我看到同事 Alex Albert 使用 Blender 渲染室内设计,我才想到可以用 Claude 为我的个人主页构建 3D 虚拟环境——尽管我以前做过电子游戏!然而,这也是我见证最多神奇瞬间发生的地方。
具体该如何操作?我通常采用以下方法:
尝试用动词替换名词: 这是来自汤姆·凯利(Tom Kelley)与乔纳森·利特曼(Jonathan Littman)著作《创新的艺术》(The Art of Innovation)中的一个经典技巧。季度业务回顾不是“一套 PPT”,而是“在 30 分钟内,让事务繁忙的人相信一个关于因果关系的故事,并在不确定性中做出决断”。提炼出动词之后,不仅能让我、也能让 Claude 更容易理解哪些领域能为这一目标带来独特且有价值的视角。
组建你的特遣队: 假设这套 PPT 必须达到极高水准,否则外星人就要毁灭地球——人类会组建一支怎样的梦之队?为了保证产出质量,我会拉入一位数据分析师、一位调查记者和一位编辑;如果需要现场演讲,或许还要加入动态视觉设计师和音频设计师;甚至还可以组建一支对抗性的“红方团队”,邀请一位做空机构研究员和一名法务会计师加入。这里很容易不小心滑向纯粹的角色扮演,这只会让上下文变得古怪而不会实质提升产出质量。使用这个方法的核心是寻找对应的学科领域与专业方法:不要让模型去“扮演”一个做空机构空头,而是让它像做空者调研仓位那样去执行具体的调研动作。
上述所有维度的雄心,都要求模型拥有可安装软件的计算机环境,以及对各类相关数据的访问权限(无论是通过 MCP 协议连接器还是其他途径)。目前已有大量工具让这成为可能,我非常建议大家配置使用。
历史早有先例
以上就是我对如何变得更有雄心的具体建议。如果你愿意多停留一分钟,我还想聊聊为什么我认为在当下这显得尤为必要。
人类自古以来就常常受困于自己对“可能发生什么”的认知定势——更糟的是,我们通常只在极其狭窄的专业领域内信任自己的能力。众所周知,米开朗基罗当年极不情愿去画西斯廷礼拜堂天顶画——因为在他看来,自己是一名雕塑家,而不是画家。
在调研关于“雄心”的课题时,我发现心理学中对此有一个专门的概念:自我效能感(Self-efficacy)。简而言之,自我效能感高的人更坚信自身的能力,从而往往能在现实中取得更大的成就。研究表明,自我效能感是可以被激发的:洛克(Locke)和拉瑟姆(Latham)深入梳理了长达 35 年的目标设定研究,得出明确结论:被设定具体且高难度目标的人,其表现显著优于那些仅仅被告知‘尽力而为’的人。Stripe 联合创始人帕特里克·科利森(Patrick Collison)长期维护着一个“快速榜单”(Fast list),记录了人类在被赋予宏伟目标和严苛期限时所完成的种种奇迹。其中一些被充分印证的思维策略包括:
通过相信事情可行去攻克难关: 1939 年,乔治·丹齐格(George Dantzig)在加州大学伯克利分校的一堂统计课上迟到了。他把黑板上的两道题抄了下来,以为那是课后作业并按时交了上去——而实际上,那是当时统计学界尚未破解的两大著名世界难题。
目睹他人做成之后去突破极限: 数十年来,人类一直试图打破“4 分钟跑完 1 英里”的极限。罗杰·班尼斯特(Roger Bannister)于 1954 年 5 月首次打破该记录,但他作为这一纪录唯一保持者的时间仅仅维持了 46 天。如果你热爱跑步,会发现类似的突破正在马拉松破 2 小时的征程中上演:塞巴斯蒂安·索维(Sabastian Sawe)在今年 4 月的伦敦马拉松赛中跑出了首个官方破 2 成绩;而在同一场比赛中,亚军也同样突破了 2 小时大关。
通过允许自己失败去成就大事: 大家似乎最喜爱的商业书籍是《身心合一的奇迹力量》(The Inner Game of Tennis,又译网球内心戏),而我上周才发现这本书。该书的核心观点是:如果你拼命去用力,你反而打不出最好的比赛状态。我们工程师或许还记得 Linux 最初的发布宣言:“这只是一项业余爱好,不会搞得像 GNU 那样庞大而专业。”
自我效能感并非新问题,但在今天它变得前所未有地关键。置身于能力每隔数周就会发生飞跃的世界,我们需要一套全新的思维与方法,让我们敢于去做那些曾经未曾考虑过、甚至不敢相信可能办到的事。
如果你发现了更多能激发雄心的新技巧,欢迎随时和我交流!
本文题图来自弗朗西斯科·戈雅(Francisco Goya)的作品《飞行方式》(Modo de volar)。
Hi, I'm Felix. I'm an eng lead working on Claude. I love watching people take flight and do things they never thought possible.
We often describe that as "being more ambitious" when using AI, which is tricky advice to follow. I wrote a tutorial:
felixrieseberg.com/how-to-be…
AI coding has gotten so powerful that I feel a bit regretful if I don't build something new every day.
#ai #llm #buildinpublic
My product development isn't limited by LLMs, coding plans, IDEs, or other tools; it's limited by my own effort and ideas.
#ai #LLM #buildinpublic
Haoriwa retweeted
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Judging by my tl there is a growing gap in understanding of AI capability.
The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.
But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.
So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.
TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.
Haoriwa retweeted
Kobe Bryant once said:
“Everyone wants to be a beast. Everyone wants to be the best. But very few people are willing to do what it actually takes. Because what it takes is boring. It is waking up at 4:00 AM. It is shooting the same shot a thousand times. It is watching the film when you are tired. People fall in love with the result, but they hate the process. You have to fall in love with the boredom. You have to fall in love with the repetition.
If you can find joy in the mundane work that no one else sees, the lights will eventually shine on you.”
Desired state: after being away from work hours, I can set aside work-related problems and bring my mood and state back into my own life, rather than still worrying about unresolved matters at work. Of course, this expectation is more directed at working in a company, rather than my own business.
你若是学习学不尽力,玩又玩不尽兴,没有扎实基础却渴望出类拔萃。看见别人勤奋就焦虑不安,自己行动时却总是“明天再说”。你不甘堕落却又不思进取,既痛恨自己的拖延,又沉迷眼前的安逸,那最后的结果一定是你什么都得不到。
- 记录自网络
Haoriwa retweeted
My first iOS app, LiLog, is officially live! It’s an offline CAN log viewer for iPhone. Feel free to give it a try! If you spot any issues, or just think it’s a piece of shit, I’d love to hear your honest feedback. Drop a comment below or shoot me an email.apps.apple.com/us/app/lilog-…










