【双语】Karpathy 最新观点：vibe coding 抬高了地板，agentic engineering 抬高了天花板
来源：https://x.com/karpathy/status/2049903821095354523
English
(Note Tweet) Fireside chat at Sequoia Ascent 2026 from a ~week ago. Some highlights:
The first theme I tried to push on is that LLMs are about a lot more than just speeding up what existed before (e.g. coding). Three examples of new horizons:
1. menugen: an app that can be fully engulfed by LLMs, with no classical code needed: input an image, output an image and an LLM can natively do the thing.
2. install .md skills instead of install .sh scripts. Why create a complex Software 1.0 bash script for e.g. installing a piece of software if you can write the installation out in words and say "just show this to your LLM". The LLM is an advanced interpreter of English and can intelligently target installation to your setup, debug everything inline, etc.
3. LLM knowledge bases as an example of something that was *impossible* with classical code because it's computation over unstructured data (knowledge) from arbitrary sources and in arbitrary formats, including simply text articles etc.
I pushed on these because in every new paradigm change, the obvious things are always in the realm of speeding up or somehow improving what existed, but here we have examples of functionality that either suddenly perhaps shouldn't even exist (1,2), or was fundamentally not possible before (3).
The second (ongoing) theme is trying to explain the pattern of jaggedness in LLMs. How it can be true that a single artifact will simultaneously 1) coherently refactor a 100,000-line code base *and* 2) tell you to walk to the car wash to wash your car. I previously wrote about the source of this as having to do with verifiability of a domain, here I expand on this as having to also do with economics because revenue/TAM dictates what the frontier labs choose to package into training data distributions during RL. You're either in the data distribution (on the rails of the RL circuits) and flying or you're off-roading in the jungle with a machete, in relative terms. Still not 100% satisfied with this, but it's an ongoing struggle to build an accurate model of LLM capabilities if you wish to practically take advantage of their power while avoiding their pitfalls, which brings me to...
Last theme is the agent-native economy. The decomposition of products and services into sensors, actuators and logic (split up across all of 1.0/2.0/3.0 computing paradigms), how we can make information maximally legible to LLMs, some words on the quickly emerging agentic engineering and its skill set, related hiring practices, etc., possibly even hints/dreams of fully neural computing handling the vast majority of computation with some help from (classical) CPU coprocessors.
中文
（推文笔记）大约一周前的 Sequoia Ascent 2026 炉边谈话要点：
我首先试图推动的主题是：大语言模型的意义远不止加速既有事物（比如编程）。三个新领域的例子：
1. menugen：一款可以完全被 LLM 吞噬的应用，无需任何经典代码：输入一张图像，输出图像——LLM 原生就能完成这件事。
2. 安装 .md 技能而非安装 .sh 脚本。为什么非要创建一个复杂的 Software 1.0 bash 脚本来安装某个软件？如果你可以直接用文字写出安装步骤，说"把这个给你的 LLM 看就行了"。LLM 是英语的高级解释器，能够智能地针对你的环境完成安装，并实时调试一切。
3. LLM 知识库——这是经典代码完全不可能实现的一个例子，因为它需要对来自任意来源、任意格式的非结构化数据（知识）进行计算，包括简单的文字文章等。
我之所以力推这些，是因为在每一个新范式变革中，显而易见的东西总是属于加速或改善既有事物，但这里我们看到的例子是：有些功能要么突然可能根本不应该存在（1、2），要么以前就根本不可能实现（3）。
第二个（进行中的）主题是试图解释 LLM 的参差不齐（jaggedness）模式。为什么同一个产物能够同时：1）连贯地重构一个 10 万行代码库，*并且* 2）告诉你走去洗车店洗车？我之前把这种来源归结为与领域的可验证性有关，这里我要补充：它还与经济学有关，因为收入/TAM（潜在市场总量）决定了前沿实验室在 RL 过程中选择将哪些内容打包进训练数据分布。你要么在数据分布中（在 RL 回路的轨道上）飞行，要么在丛林里用砍刀开路。仍不完全满意，但这是一场持续的战斗——如果你想在实践中利用 LLM 的力量同时避免其陷阱，就需要建立一个准确的 LLM 能力模型，这也让我想到……
最后一个主题是原生 agent 经济。产品和服务分解为传感器、执行器和逻辑（分散在 1.0/2.0/3.0 所有计算范式中），如何让信息对 LLM 最大限度地清晰可读，对快速兴起的 agentic engineering 及其技能集、相关招聘实践等也有一些思考，甚至可能涉及全神经计算处理绝大部分计算、辅以（经典）CPU 协处理器的展望/梦想。