How to Make Sense of AI

This is Part 1 on a short series on sensemaking.这是关于意义建构的短篇系列的第一部分。

It is 2026 and AI hype is everywhere.现在是2026年，AI炒作无处不在。

If you’re like most people, you’re probably feeling some fear that you’re falling behind. Perhaps your fellow company operators are talking about successful AI use in their companies, and you’re questioning if you need to retool everything. Perhaps your friends are freaking out about losing their jobs. Perhaps you’re spending countless cycles trying to predict what’s coming next.如果你和大多数人一样，你可能正在感到某种恐惧，担心自己落后了。也许你的公司同行正在谈论他们公司中成功的AI应用，而你在质疑是否需要彻底改造一切。也许你的朋友们正在为失业而恐慌。也许你正花费无数精力试图预测接下来会发生什么。

This is understandable. Widespread panics are more common during technological revolutions. It feels scary when things are changing so quickly — and in ways that will impact your livelihood and therefore your life. If there’s anything that we’ve learnt during the Covid years, it’s that humans don’t like uncertainty. Who knows what the landscape of work would look like in a few short years? Nobody does.这是可以理解的。在技术革命期间，广泛的恐慌更为常见。当事情变化如此之快——而且是以影响你生计、因此影响你生活的方式变化时——感到害怕是正常的。如果说新冠疫情这几年我们学到了什么，那就是人类不喜欢不确定性。短短几年后工作的面貌会是什么样子？没有人知道。

And yet there is an effective way to make sense of these accelerating changes. With the right frame, you can maintain your equanimity and focus on the right things — that is, only the things that may affect your outcomes. This doesn’t mean sitting back and passively observing the changes around you. In some ways you’re going to take more action as you investigate the new capabilities of this technology. But you want to be able to investigate without flailing around.然而，有一种有效的方法来理解这些加速的变化。有了正确的框架，你可以保持镇定，专注于正确的事情——也就是说，只关注那些可能影响你结果的事情。这并不意味着袖手旁观，被动地观察周围的变化。在某些方面，当你调查这项技术的新的能力时，你会采取更多行动。但你希望能够有条不紊地调查，而不是手忙脚乱。

The ideal response looks like this: you are able to make decisions without being emotionally affected, without feeling FOMO, and without the distraction and panic that has claimed so many in the business world. You will seem oddly quiet and determined, unfazed by any change that comes your way. You will ignore unfounded AI doomerism and unfounded hype equally; you are able to test assertions and measure outcomes without emotion. Done right, this frame will help you become effective at ‘fast adaptation under uncertainty’, so that you know how to direct your attention and therefore your actions.理想的反应是这样的：你能够做出决定，而不受情绪影响，不感到错失恐惧（FOMO），也不被那种已经吞噬了商界许多人的分心和恐慌所困扰。你会显得异常安静而坚定，对任何迎面而来的变化都毫不动摇。你会同样无视毫无根据的AI末日论和毫无根据的炒作；你能够不带情绪地验证断言和衡量结果。做得好，这个框架将帮助你有效地进行'不确定性下的快速适应'，从而知道如何引导你的注意力，进而引导你的行动。

In 2023, at the start of the current AI boom, I decided to read a bunch of histories about previous technological bubbles revolutions. The most helpful book was Engines That Move Markets, by Alisdair Nairn, which covers 10 different tech bubbles. A second book that was helpful (h/t David MacIver) was David Edgerton’s The Shock of the Old, which I’ve mentioned in passing in Letter to a Young Person Worried About AI. I also reread multiple biographies that coincided with certain new technologies — Reminiscences of a Stock Operator was one (the protagonist, Jesse Livermore, dealt a fair bit with railway stocks), plus bits from Titan, Ron Chernow’s history of John D Rockefeller (oil) and chunks of The House of Morgan, Chernow’s history of J. P. Morgan (railroads, steel, and oil). Then I skimmed various biographies of the early PC era, just to get a feel for the takeoff stages of these bubbles.2023年，在当前AI热潮开始时，我决定阅读大量关于以往技术泡沫革命的历史。最有帮助的书是Alisdair Nairn的《Engines That Move Markets》，涵盖了10个不同的技术泡沫。第二本有帮助的书（感谢David MacIver推荐）是David Edgerton的《The Shock of the Old》，我在《致担忧AI的年轻人的一封信》中顺便提到过。我还重读了与某些新技术同时期的多本传记——《Reminiscences of a Stock Operator》是其中之一（主角Jesse Livermore大量涉及铁路股票），还有Ron Chernow的《Titan：约翰·D·洛克菲勒的历史》（石油）和《The House of Morgan》（Chernow关于J.P.摩根的历史，涉及铁路、钢铁和石油）的部分内容。然后我浏览了早期PC时代的各种传记，只是为了感受这些泡沫起飞阶段的情况。

All told, my reading program took 1.5 years. I couldn’t articulate what I was looking for then, but I think I can articulate it now. First, I wanted to know what it felt like as these technological revolutions progressed. I was too young for the dotcom bust, though I watched the outcome of that revolution unfold in my teens and my 20s. (One common thing with all of these revolutions, by the way, is that they take multiple decades to unfold). The second thing I was looking for? I wanted a generalised strategy for sensemaking revolutionary new technology.总的来说，我的阅读计划花了1.5年。当时我无法说清楚自己在寻找什么，但现在我想我能够说清楚了。首先，我想知道随着这些技术革命的推进，当时是什么感觉。对于互联网泡沫破灭，我当时太年轻了，尽管我在十几岁和二十多岁时目睹了那场革命的结果。（顺便说一下，所有这些革命的一个共同点是，它们需要几十年的时间才能展开）。我想寻找的第二件事是什么？我想要一个用于理解革命性新技术的通用策略。

This essay will lay out what I’ve worked out for myself. It will be short. It contains a method of sensemaking that you may adapt to your own circumstances, assuming that you are a business operator. (Note: if you are an investor, this essay is not for you. The sensemaking needs of investors are significantly more advanced). The sensemaking approach outlined here is generally useful; it applies to any new technology. This means that once you master this approach, you will be able to apply it to all other paradigm-changing technologies that will emerge over the course of your life.这篇文章将阐述我自己摸索出来的东西。它会很短。它包含一种意义建构方法，你可以根据自己的情况进行调整，假设你是一名企业经营者。（注意：如果你是投资者，这篇文章不适合你。投资者的意义建构需求要复杂得多）。这里概述的意义建构方法具有普遍实用性；它适用于任何新技术。这意味着一旦你掌握了这种方法，你就能够将其应用于你一生中出现的所有其他改变范式的技术。

In latter instalments, we will cover what sensemaking actually is, and how experts sensemake more effectively compared to novices. This is informed by research funded by the US Military decades ago, which was in turn motivated by a need to help warfighters and intelligence analysts make sense of uncertain, fast-changing scenarios. Then, with some theory at hand, we will examine why the method laid out in this essay is a good start, but is actually not enough.在后续部分中，我们将介绍意义建构究竟是什么，以及专家与新手相比如何更有效地进行意义建构。这是基于几十年前美国军方资助的研究，其动机是帮助作战人员和情报分析师理解不确定、快速变化的场景。然后，有了一些理论在手，我们将审视为什么这篇文章中提出的方法是一个好的开始，但实际上还不够。

The Method方法

First, let’s lay down a few ground rules.首先，让我们确立几条基本规则。

The first ground rule is that your attention is a limited resource. You are inundated with news, opinions, Substack takes, and (god forbid) tweets, which are designed to elicit responses from you. Some of these responses will be helpful. Others will not.第一条基本规则是，你的注意力是一种有限的资源。你被新闻、观点、Substack文章，以及（天哪）推文所淹没，它们旨在引发你的反应。其中一些反应会有帮助。另一些则不会。

At its core, sensemaking is the art of regulating attention. This is a fancier way of saying that you must know what to ignore. And then you must have the discipline of mind to ignore those things, so that you may focus only on things that will help you.从根本上说，意义建构是调节注意力的艺术。这是一种更花哨的说法，意思是你必须知道该忽略什么。然后你必须有心理上的纪律去忽略那些东西，这样你才能只专注于对你有帮助的事情。

Both steps are difficult in their own ways, but this piece will focus only on the first step. The second step — actually sticking with this approach — is an exercise for the alert reader.这两个步骤各有各的困难，但这篇文章只关注第一步。第二步——真正坚持这种方法——留给细心的读者作为练习。

A second ground rule is the concept of ‘Outcome Orientation’. We’ve talked about putting Outcome Orientation to practice on Commoncog before — the idea is simple but remarkably effective:第二条基本规则是'结果导向'的概念。我们之前在Commoncog上讨论过将结果导向付诸实践——这个想法简单但 remarkably 有效：

At all times, whenever you are doing something or reading something, you should ask yourself the question:

“What is the outcome I am trying to achieve here?”

You may then continue with the action or consumption if you wish, but you must answer the question honestly first.在任何时候，当你在做某事或阅读某物时，你都应该问自己这个问题：'我在这里试图实现的结果是什么？'然后你可以继续进行该行动或消费，但你必须先诚实地回答这个问题。

Applying Outcome Orientation to all your information consumption practices will cause something magical to happen: you will no longer feel information overload. At this point enough folks in Commoncog’s members community have applied it to their lives that I can say this with some confidence: simply noticing where your attention is going will change the way you allocate that attention.将结果导向应用于你所有的信息消费实践，将会发生神奇的事情：你将不再感到信息过载。到目前为止，Commoncog会员社区中已有足够多的人将其应用于他们的生活，使我可以相当有把握地说：仅仅注意到你的注意力去向何处，就会改变你分配注意力的方式。

That’s it, just two ground rules. Now let’s talk about the method directly.就这些，只有两条基本规则。现在让我们直接谈谈方法。

The basic method, simply stated, is as follows:基本方法，简单地说，如下：

1. You must ignore all opinions, analysis, predictions, fanciful essays ‘from the future’, ‘situational awareness updates’ and scenario forecasts about AI. It doesn't matter how eloquent they are, how smart these authors seem, what seat they have, or whether their assessment of AI is compelling — you should ignore all of it.你必须忽略所有关于AI的观点、分析、预测、来自未来的奇幻文章、'情境意识更新'和情景预测。不管它们多么雄辩，作者看起来多么聪明，他们坐什么位置，或者他们对AI的评估多么有说服力——你应该全部忽略。
2. You will pay attention only to detailed field reports of use. This may take on any form: tweets, YouTube videos, screencasts, blog posts. A field report is acceptable only if it is adequately detailed, but you should also take into account who the author is, what their context of use might be, and what they are trying to accomplish. If the piece contains opinions or forecasts alongside the field report, you will pay attention only to the field report, and ignore the more speculative / opinionated bits. You will not take those subjective bits seriously — the right model here is that you should treat it like the mutterings of a friend who is high on LSD.你只关注详细的使用现场报告。这可以采取任何形式：推文、YouTube视频、屏幕录制、博客文章。现场报告只有在足够详细的情况下才可接受，但你也应该考虑作者是谁，他们的使用背景可能是什么，以及他们试图实现什么。如果文章中包含与现场报告一起的观点或预测，你只关注现场报告，忽略更具推测性/主观性的部分。你不会把那些主观部分当真——这里的正确模型是，你应该把它当作一个吸了LSD的朋友的喃喃自语。
3. Whilst paying attention to detailed field reports of use you are looking to answer the four questions of uncertainty. The four questions are: (a) What new outcomes are suggested by this field report? (b) What are some actions I may take in response to it? (c) What are the relative values of these possible outcomes to me (given who I am, what my company does, what I value, and what my goals are)? (d) What are the causal relationships here?在关注详细的使用现场报告时，你试图回答关于不确定性的四个问题。这四个问题是：(a) 这个现场报告暗示了什么新结果？(b) 我可以采取哪些行动来应对它？(c) 这些可能的结果对我的相对价值是什么（鉴于我是谁、我的公司做什么、我重视什么以及我的目标是什么）？(d) 这里的因果关系是什么？

That’s it.就这些。

In truth, the important thing to focus on are the four questions in Step 3. When you are operating in your career or in your life, you are not actually interested in “the impact of new technology X on society” or “the impact of AI on job loss.” You are mostly interested in the impact of AI on your career, and your life outcomes (and of course the life outcomes of those you love).事实上，重要的是关注第三步中的四个问题。当你在职业生涯或生活中运作时，你实际上并不关心'新技术X对社会的影响'或'AI对失业的影响'。你主要关心的是AI对你职业生涯的影响，以及你的生活结果（当然还有你所爱的人的生活结果）。

You do not need to hold opinions on those broader questions in order to be effective.你不需要对这些更广泛的问题持有观点才能有效。

Seeking the answers to these four questions will force you to sensemake for your specific situation. The failure mode that you want to prevent is speculating about useless things that feel like productive forecasting (but actually aren’t, because they aren’t directly relevant to your context).寻求这四个问题的答案将迫使你在你的具体情况下进行意义建构。你想要防止的失败模式是对无用的事情进行推测，这些事情感觉像是富有成效的预测（但实际上不是，因为它们与你的背景没有直接关系）。

Here are some example answers to the four questions:以下是四个问题的一些示例答案：

1. What are the possible outcomes here? On September 29 2025, Microsoft Deputy CTO Sam Schillace published I Have Seen The Compounding Teams, which describes observing ‘two or three teams’ that produce working, usable software at a high rate of output, without a single line of human-written code and without human code review. He linked to an open source repository by Microsoft Research called ‘Amplifier’ that aims to accomplish this outcome, and reported that it takes around six months of harness improvement to get to this point. On February 11 2026, OpenAI published Harness engineering: leveraging Codex in an agent-first world. This report is a little more suspect given that the author’s organisation — OpenAI — has a vested interest in keeping the AI hype cycle going. Nevertheless there was enough detail in the field report to be useful for our purposes (again: around possible outcomes). On March 16 2026, Schillace published The Rise of Taste — reporting that he had successfully used multiple dev machines running autonomously for a few days each to make a) ‘a high fidelity clone of Microsoft Word using web technologies’, b) ‘OpenClaw but for Microsoft 365’ and c) a ‘security filter product for agents that is also meant for enterprise.’ Each of these applications were produced without a single line of human-written code and with minimal code review. So, a possible outcome here is that ‘it is possible to produce complex, usable software at high velocity without any manual human intervention, but it requires around six months of building scaffolding for the AI agents’.这里可能的结果是什么？2025年9月29日，微软副首席技术官Sam Schillace发表了《I Have Seen The Compounding Teams》，描述了观察到'两三个团队'以高产出率生产可用的工作软件，没有一行人工编写的代码，也没有人工代码审查。他链接了微软研究院名为'Amplifier'的开源仓库，旨在实现这一结果，并报告说大约需要六个月的工具改进才能达到这一点。2026年2月11日，OpenAI发表了《Harness engineering: leveraging Codex in an agent-first world》。这份报告更值得怀疑，因为作者所在的组织——OpenAI——有维持AI炒作周期的既得利益。尽管如此，现场报告中有足够的细节对我们的目的有用（再次强调：关于可能的结果）。2026年3月16日，Schillace发表了《The Rise of Taste》——报告说他已成功使用多台开发机器自主运行数天，分别制作了a）'使用网络技术的高保真Microsoft Word克隆版'，b）'适用于Microsoft 365的OpenClaw'，以及c）'面向企业的代理安全过滤产品'。这些应用程序中的每一个都是在没有一行人工编写的代码和最少代码审查的情况下生产的。因此，这里的一个可能结果是'可以在没有任何人工干预的情况下以高速度生产复杂、可用的软件，但需要大约六个月的时间来构建AI代理的脚手架'。
2. What are the further actions I may take? In response to Vaughn Tan’s field report about his ‘boring, tiny tools’ (or BTTs) and Craig Mod’s field report about building accounting software for his idiosyncratic tax situation: I conclude that I may use agentic coding tools to build BTTs in order to help me solve repetitious, brain-dead tasks in my business and in my life. Furthermore, I know that experimenting with these tools in service of building BTTs will take me no more than a few days of work for each tool.我可以采取哪些进一步行动？针对Vaughn Tan关于他的'无聊、微小工具'（或BTT）的现场报告，以及Craig Mod关于为他的特殊税务情况构建会计软件的现场报告：我得出结论，我可以使用代理编码工具来构建BTT，以帮助我解决业务和生活中重复的、无脑的任务。此外，我知道为了构建BTT而使用这些工具进行实验，每个工具只需要几天的工作。
3. What are the relative value of outcomes to me? As a business owner, the value of being able to produce BTTs in my spare time is that I can reduce friction in certain parts of the business. (To be clear: I’ve already done this, so there’s no need to verify). However, the value of ‘compounding teams’ shipping large chunks of complex software with only a handful of engineers is a much higher value outcome. The question is if it’s worth it (for my specific business) to invest six months into custom infrastructure, tooling, and process experimentation to achieve this outcome, or if it’s better to wait for commercial versions of Amplifier to be released. I know, from my network, that various startups in Silicon Valley are all aiming for that outcome. A more important action might be to find folks in these teams and develop relationships with them, so that I may keep tabs on their discoveries.这些结果对我的相对价值是什么？作为企业主，能够在业余时间制作BTT的价值在于，我可以减少业务某些部分的摩擦。（明确地说：我已经这样做了，所以不需要验证）。然而，'复合团队'仅用少数工程师就能交付大量复杂软件的价值是一个更高价值的结果。问题是，对于我的特定业务来说，投资六个月来构建定制基础设施、工具和流程实验以实现这一结果是否值得，还是等待Amplifier的商业版本发布更好。我从我的网络中得知，硅谷的各种初创公司都在瞄准这一目标。一个更重要的行动可能是找到这些团队中的人并与他们建立关系，以便跟踪他们的发现。
4. What are the causal relationships here? All the best practices for software engineering — test-driven development, blue-green deployments, continuous delivery with high cardinality, high dimensionality observability — turn out to be useful (even necessary) for higher velocity, AI-first software production.这里的因果关系是什么？所有软件工程的最佳实践——测试驱动开发、蓝绿部署、具有高基数、高维度可观测性的持续交付——结果证明对于更高速度的AI优先软件生产是有用的（甚至是必要的）。

Naturally, seeking answers to these four questions should cause you to change your behaviour:自然地，寻求这四个问题的答案应该会导致你改变行为：

1. You may set up a Slack channel or a WhatsApp group to sensemake collectively. Brief your friends or colleagues on the logic for field reports, and then get them to share links to reports with the implicit understanding that everyone is trying to answer the four questions of uncertainty for their specific context. Commoncog’s private members forum has a thread set up specifically for this purpose; I make it a point to call out predictions as — basically — science fiction.你可以设置一个Slack频道或WhatsApp群组来集体进行意义建构。向你的朋友或同事简要说明现场报告的逻辑，然后让他们分享报告链接，并默认理解每个人都在试图为他们的具体背景回答关于不确定性的四个问题。Commoncog的私人会员论坛为此专门设立了一个线程；我会特意指出预测基本上是科幻小说。
2. You may want to take action to verify that various bits of information in the field reports are true. For instance, building an autonomous coding harness might take too much time, but a cheaper experiment is to clone a major open source project, from scratch, with minimal human oversight, simply by using the existing test suite.你可能想要采取行动来验证现场报告中的各种信息是否属实。例如，构建自主编码工具可能花费太多时间，但更便宜的实验是克隆一个主要的开源项目，从头开始，只需最少的人工监督，仅使用现有的测试套件。
3. You may also want to take action to verify that the various outcomes are a good fit for your situation. For instance, finding gaps in your existing life and then vibe-coding boring tiny tools in response to those gaps is a fairly cheap experiment to do, and will give you lots of context-specific information about how this new technology works in your situation.你可能还想采取行动来验证各种结果是否适合你的情况。例如，找到你现有生活中的空白，然后针对这些空白进行vibe-coding无聊的小工具，这是一个相当便宜的实验，会给你很多关于这项新技术在你情况下如何运作的特定情境信息。

In your downtime, you may speculate about the potential impact of all of this technological change, but you will not allow it to affect your sensemaking actions.在你的空闲时间，你可以推测所有这些技术变革的潜在影响，但你不会允许它影响你的意义建构行动。

Why Does This Work?为什么这有效？

A couple of years ago I observed that effective businesspeople do not ‘predict the future’ more successfully than less effective businesspeople. It is simply too difficult and too cognitively expensive to accurately (or reliably) predict the future. Instead, the best businesspeople do something else: they do fast adaptation under uncertainty.几年前我观察到，高效的企业经营者并不比低效的企业经营者更成功地'预测未来'。准确（或可靠）预测未来实在太困难、认知成本太高。相反，最好的企业经营者做另一件事：他们在不确定性下快速适应。

This requires a different stance, and a different set of skills. For starters, forecasters tend to want to be right; businesspeople don’t care about being right, they just want to win. This stance means that the business should be set up for experimentation and information sharing. It implies that the business must be agile enough: it should be able to change directions in response to new information. It implies that advantage accrues to those who are able to sensemake more effectively than their competitors.这需要不同的立场和不同的技能组合。首先，预测者往往想要正确；企业经营者不在乎是否正确，他们只想赢。这种立场意味着企业应该为实验和信息共享而设立。它意味着企业必须足够敏捷：它应该能够根据新信息改变方向。它意味着优势会累积给那些比竞争对手更有效地进行意义建构的人。

Another way to put this — a pithier way — is that you don’t have to predict the future if you can see the present moment clearly (and of course, that you can act in response to it).另一种更简洁的说法是，如果你能看清当下（当然，并且能据此采取行动），你就不必预测未来。

When a new technology emerges, superior sensemaking is usually about:当新技术出现时，更优秀的意义建构通常关乎：

• What new outcomes are now possible thanks to this new technology? (Notice that this is not necessarily directly due to the technology; the emergence of the technology may cause some other change in society that you may exploit).由于这项新技术，现在什么新结果成为可能？（注意，这不一定直接由技术导致；技术的出现可能会引起社会的某些其他变化，你可以利用这些变化）。
• What actions can we take now that we have the new technology? (In slightly more technical language: what new affordances does this give us?)现在我们有了新技术，可以采取什么行动？（用稍微更技术性的语言说：这给了我们什么新的 affordances？）
• What are the relative value of outcomes to us, given our business and our competitive position? Is there something here that we can leverage differently from our competitors?鉴于我们的业务和竞争地位，这些结果对我们的相对价值是什么？这里有什么我们可以与竞争对手不同地利用的东西吗？
• What are the causal relationships? Are there new levers we can pull that — if we do it quietly or secretly — would result in an edge against our competitors?因果关系是什么？是否有我们可以拉动的新杠杆——如果我们悄悄或秘密地做——会在与竞争对手的对抗中带来优势？

Why ignore analysis, opinions, or forecasts, though? The answer is straightforward: when a new technology emerges, impacts are uneven and ultimately unpredictable. When the internet first arrived, say at the peak of the dotcom bubble in 1999, no one could’ve predicted that taxi drivers would be threatened. You are better served keeping a paranoid stance and observing how the technology affects your specific context more than you will be by reading the analysis of those who sit in different parts of the economy from you.为什么要忽略分析、观点或预测呢？答案很直接：当新技术出现时，影响是不均衡的，最终是不可预测的。当互联网刚出现时，比如在1999年互联网泡沫的顶峰，没有人能预测到出租车司机将受到威胁。你最好保持偏执的立场，观察技术如何影响你的具体背景，而不是阅读那些坐在经济不同位置的人的分析。

(This is, by the way, the meta-lesson of Only The Paranoid Survive — when the Internet arrived, legendary Intel CEO Andy Grove wasn’t sitting around reading pundits. He was periodically redoing Porter’s Five Forces Analysis on the Internet’s impact on Intel — and you can betcha he was basing it on actual capability reports, not opinion columns.)（顺便说一下，这是《Only The Paranoid Survive》的元教训——当互联网到来时，传奇的英特尔CEO安迪·格鲁夫并没有坐在那里读专家评论。他定期重做波特五力分析，评估互联网对英特尔的影响——而且你可以打赌，他是基于实际的能力报告，而不是观点专栏。）

This leads us to our second point: during times of upheaval, the attention economy creates large incentives for people to prey on fears in order to advance their followings. You should expect to see plenty of content that are perfectly tuned to go viral. All such content is for the author’s benefit, not yours. Over the past few weeks, various people in my circles have been overtaken by fear in response to viral essays, only to recover their senses after a few days (and with the subsequent revelation that the author was not believable). You can skip all this if you treat all AI takes equally: like trash.这引出了我们的第二点：在动荡时期，注意力经济创造了巨大的激励，让人们利用恐惧来扩大自己的追随者。你应该会看到大量为病毒式传播而精心调校的内容。所有这些内容都是为了作者的利益，而不是你的。在过去几周里，我圈子里的各种人被病毒式文章所引发的恐惧所吞噬，几天后才恢复理智（并随后发现作者不可信）。如果你平等对待所有AI观点：像垃圾一样，你就可以跳过所有这些。

Third, many of the opinions produced in response to a ‘terrifying’ new technology are produced for self-soothing reasons — they serve no purpose other than to comfort the author (and the people who share them on social channels). Witness the sheer number of “woe, programming has forever changed” articles that have emerged in the first quarter of 2026. Some of these reactions are embedded in otherwise informative field reports. I’ve learnt to tune those out, so that I can form my own opinions on the outcomes that I care about.第三，许多针对'可怕'新技术产生的观点是出于自我安慰的原因——它们除了安慰作者（以及在社交渠道上分享它们的人）之外没有任何目的。看看2026年第一季度涌现的大量'呜呼，编程永远改变了'的文章。其中一些反应嵌入在原本信息丰富的现场报告中。我学会了忽略那些，以便对我关心的结果形成自己的看法。

Finally, reading opinion pieces is just too time consuming. There are simply too many takes out there; prognosticators tend to like to hear themselves talk. There are always more people who have an opinion about a thing than those who are willing to actually do things. (And, yes, creating a nonsense simulation of AI impact and then writing it up is equivalent to having an opinion about a thing. It is not equivalent to reporting from doing — that is, testing against reality.)最后，阅读观点文章太耗时了。外面的观点实在太多；预言家们总是喜欢听自己说话。对某事有意见的人总是比愿意实际去做的人多。（而且，是的，创建一个AI影响的无意义模拟然后写下来，等同于对某事有意见。这并不等同于从实践中报告——即，与现实进行测试。）

The truth is that you must dedicate some time to experimentation to verify the field reports that you find. This implies that reading other people’s opinions has a real opportunity cost. The information you generate from experimentation is some of the most valuable you will get. If nothing else, it lets you know the capabilities you must build in order to adapt.事实是，你必须投入一些时间进行实验，以验证你发现的现场报告。这意味着阅读他人的观点有真正的机会成本。你从实验中生成的信息是你将获得的最有价值的信息之一。如果不说别的，它让你知道为了适应你必须建立的能力。

In the end, all of this may be reduced back down to the four questions of uncertainty. When reading a piece, you should always ask yourself: “which of the four questions does this piece of content answer for me?” If the answer is “none of them”, then you should ask yourself: “why am I still reading it?”最终，所有这一切都可以归结为关于不确定性的四个问题。在阅读一篇文章时，你应该始终问自己：'这篇文章为我回答了四个问题中的哪一个？'如果答案是'没有一个'，那么你应该问自己：'我为什么还在读它？'

You are allowed to say “because I want to soothe myself” or “because I want entertainment” and continue with the pieces. But you must be honest with yourself.你可以说'因为我想安慰自己'或'因为我想娱乐'，然后继续阅读这些文章。但你必须对自己诚实。

How to Start Putting This to Practice如何开始付诸实践

Notice what I’m saying, by the way: consuming opinion pieces is NOT necessarily a waste of time in other spheres of life. For instance, I am not believable about geopolitics or war; like you, I read analysis in order to make sense of what’s going on in the Middle East. Of course, whether one should spend much time reading about geopolitics is an exercise for the alert reader. (What outcomes are you trying to accomplish there?)注意我在说什么，顺便说一下：消费观点文章在其他生活领域不一定是浪费时间。例如，我在地缘政治或战争方面没有可信度；和你一样，我阅读分析是为了理解中东正在发生什么。当然，一个人是否应该花大量时间阅读地缘政治，这是留给细心读者的练习。（你在那里试图实现什么结果？）

The method presented in this essay recommends against reading takes because it is not as useful when dealing with an uncertain new technology — one that will likely impact your life directly. But because reading analysis is such a normal thing in other spheres of life, it is important to start with this if you want to put the method to practice.这篇文章中提出的方法建议不要阅读观点，因为当处理一个不确定的新技术时——一个可能直接影响你生活的技术——这样做不太有用。但因为阅读分析在其他生活领域是如此正常的事情，如果你想将方法付诸实践，从这里开始很重要。

The first step to using this method is to go cold turkey on reading analysis. This gives you the mental space (and the time) to do the other sensemaking activities described in this essay.使用这种方法的第一步是彻底戒掉阅读分析。这给你心理空间（和时间）去做这篇文章中描述的其他意义建构活动。

Start now. Spend a week just rolling your eyes at AI prognostication. Treat predictions and opinions with the same level of respect you would the babblings of a small child. (That is: respectfully, but not seriously). Overcompensate on this to compensate for your natural desire to consume opinions. The way I do this is that I imagine myself discarding opinions in conversations about AI; I remind myself to ask detailed questions of use.现在就开始。花一周时间对AI预言翻白眼。以对待小孩咿呀学语的同等尊重程度对待预测和观点。（即：有礼貌地，但不当真）。在这方面过度补偿，以抵消你消费观点的自然欲望。我这样做的方式是，想象自己在关于AI的对话中摒弃观点；我提醒自己询问关于使用的详细问题。

After you’ve fixed this tendency to consume opinions, the next step is relatively straightforward: ruthlessly ask the four questions of uncertainty every time you read something about AI.在你纠正了消费观点的倾向之后，下一步相对简单：每次阅读关于AI的内容时，无情地问关于不确定性的四个问题。

At the end of the day, it is the four questions that determine what you will do.归根结底，是四个问题决定了你会做什么。

Wrapping Up总结

While the approach presented in this essay is a useful foundation for sensemaking changes in AI, it is not enough. In the next instalment, we will talk about the most useful theory for sensemaking that is currently known, so that we may identify problems with this approach. Then, after we are equipped with more precise language, we will talk about how to adapt this approach to be more effective.虽然这篇文章中提出的方法对于理解AI的变化是一个有用的基础，但这还不够。在下一部分中，我们将讨论目前已知的最有用的意义建构理论，以便我们识别这种方法的问题。然后，在我们掌握了更精确的语言之后，我们将讨论如何调整这种方法使其更有效。

The four questions of uncertainty in this essay is from author, consultant and business academic Vaughn Tan’s Not Knowing series. He developed this framework over the course of 2023 and 2024 to help organisations deal better with uncertainty.这篇文章中的关于不确定性的四个问题来自作家、顾问和商业学者Vaughn Tan的《Not Knowing》系列。他在2023年和2024年期间开发了这个框架，以帮助组织更好地应对不确定性。

While I am not available for consulting, Vaughn is. He runs workshops and training programs for various institutions. Perhaps you’d like him to improve the sensemaking of your organisation? You may contact him here.虽然我不提供咨询，但Vaughn提供。他为各种机构举办工作坊和培训项目。也许你想让他提升你组织的意义建构能力？你可以在这里联系他。

This is Part 1 in a short series on sensemaking. You may read Part 2 here: How Experts Sensemake.这是关于意义建构的短篇系列的第一部分。你可以在这里阅读第二部分：How Experts Sensemake。

Originally published 18 March 2026, last updated 04 May 2026.最初发表于2026年3月18日，最后更新于2026年5月4日。