搜索 (search)
集里怎么说它
- 《AI 搜索量今年将超过人类:Exa 的完美搜索宣言》(00:20起):本集说搜索生态正在被 AI 改写:AI 系统的每日搜索量今年预计超过人类,每一次软件交互都以搜索为根基;完美搜索不是单一产品,而是为 5000 个客户建的 5000 个搜索引擎。
① 提到它的金句
30 条
Webflow,我们看到 LLM 流量和 Google 搜索流量之间的转化率差异是 6 倍。
指向原始笔记的链接
Webflow, we saw a 6X conversion rate difference between LLM traffic and Google Search traffic.
—— Ethan Smith · [14:47]
如果 Google 是一个针对 ChatGPT 回答的搜索引擎,Google 就没有理由存在了。
指向原始笔记的链接
If Google’s a search engine for ChatGPT responses, there’s no reason for Google to exist.
—— Ethan Smith · [55:57]
一个没有人策划的大脑会变成垃圾场,优秀的搜索检索会以绝对的信心检索出一个陈旧的事实。
指向原始笔记的链接
A brain nobody curates becomes a garbage dump with great search retrieval will surface a stale fact with total confidence.
—— Garry Tan · [14:30]
因此,将这种推理嵌入到我们的搜索智能体中,实际上让我们能够将准确率从大约 50% 提高到了 90。
指向原始笔记的链接
So embedding this sort of reasoning into our search agent is actually something that got us up from roughly like 50% accuracy all the way to 90.
—— 嘉宾 · [11:53]
虽然 AI 对我们所有人来说是一个极其有帮助的工具,就像互联网搜索一样,但我仍然认为炒作尚未被现实追上。
指向原始笔记的链接
While AI is an extremely helpful tool to all of us in the same way that, you know, internet search was, I still think the hype is outpacing the reality quite yet.
—— Julie Bort · [29:43]
一个没有人整理的大脑就是一个拥有伟大搜索功能的垃圾场。
指向原始笔记的链接
A brain nobody curates is a garbage dump with great search.
—— Garry Tan · [23:37]
我想事实可能是,今天使用 AI 的人里 99% 只是把它当作一个搜索引擎。
指向原始笔记的链接
I think the fact is probably 99% of people that use AI today just use it as a search engine.
—— Adam · [52:49]
我们在 Parallel 的观点是,人类点击数据是一个 bug,使用搜索进行工作的智能体应该依赖智能体反馈,而不是人类反馈。
指向原始笔记的链接
Our view at Parallel is that human click data is a bug and agent doing work with search should rely on agent feedback, not human feedback.
—— Parag · [00:00]
如果你使用 parallel 搜索,你在很大程度上会在你的智能体中使用不到一半的 token。
指向原始笔记的链接
If you use parallel search, you will, for the most part, use under half the tokens in your agent.
—— Parag · [13:49]
我们在 Parallel 的观点是,人类点击数据是一个缺陷,进行搜索工作的智能体应该依赖智能体反馈,而不是人类反馈。
指向原始笔记的链接
Our view at Parallel is that human click data is a bug and agent doing work with search should rely on agent feedback, not human feedback.
—— Parag · [00:00]
所以我们说,与搜索引擎在第一天相比,坐在搜索引擎上的人类似乎更容易竞争。
指向原始笔记的链接
So we said, it seems like humans sitting on search engines are far way easier to compete with than a search engine on day zero.
—— Parag · [08:43]
我们可以看到,在许多不同的模型之间,与试图搜索网页、找到正确的工具相比,能节省近 50% 的 token,不用从你的代码库中拉取 markdown 文件。
指向原始笔记的链接
We can see that between many different models, almost 50% token saving on instead of trying to search the web, find the right tools, pulling the markdown files from your code base.
—— Christopher Burns · [11:22]
大多数 LLM 系统本质上就是一个搜索问题。
指向原始笔记的链接
Most LLM systems are just a search problem.
—— Jean-Denis Greze · [01:22]
一个是我们会有泛化能力更好的系统,因为每当你增加一个搜索引擎、或向量数据库(它就是一种搜索引擎)、或某种步骤间的路由逻辑,这种人工整理就真的限制了系统做很多不同事情的能力。
指向原始笔记的链接
One is we will have systems that can generalize better because every time you add a search engine or a vector database, which is a search engine, or some conditional logic to route between the steps, this human curation really limits the ability of that system to do a lot of different things.
—— Alexander Whedon · [30:24]
你想想,慢速搜索是没有市场的。
指向原始笔记的链接
If you think about it, there’s no market for slow search.
—— Andrew Feldman · [53:02]
几百毫秒的写延迟。如果你像 Shopify 那样为名人闪购做库存预留,那可不行。但对搜索来说非常非常好。
指向原始笔记的链接
Hundreds of milliseconds. If you’re like Shopify and doing inventory reservations for a Kylie Jenner flash sale, not gonna work. Very, very good for search.
—— Simon Eskildsen · [08:17]
在网络规模下,文本搜索比做向量搜索更困难、计算成本更高。
指向原始笔记的链接
Text search at web scale is more difficult and more computationally expensive than doing vector search.
—— Simon Eskildsen · [19:06]
问题是那样每次搜索要花,比如,1000 万美元。
指向原始笔记的链接
The problem is that would cost, like, $10 million per search.
—— Will Bryk · [06:24]
我们只是想做一个更好的 Google 搜索,以后再想办法怎么赚钱。
指向原始笔记的链接
We were just trying to build a better search on Google. We’ll figure out how to make money later.
—— Will Bryk · [08:42]
如果对人类最优的搜索引擎对 AI 系统同样最优,那就太疯狂了。
指向原始笔记的链接
And it would be crazy if the same search engine that was optimal for humans was also optimal for these AI systems.
—— Will Bryk · [09:46]
实际上是我们为 5000 个客户中的每一个都建了 5000 个搜索引擎,对吧?
指向原始笔记的链接
It’s actually we have 5,000 search engines for each of our 5,000 customers, right?
—— Will Bryk · [12:52]
我们在数十亿个预先建好索引的视频中进行搜索,不是按关键词而是按动作搜索,并为你提供可以直接使用的片段。
指向原始笔记的链接
We search through billions and billions of pre-indexed videos, not by keywords but by actions, and we provide you with ready-to-use clips.
—— Rafael Levi · [09:32]
所以你本质上是在给网页搜索分配计算量,以便在模型上节省计算量。
指向原始笔记的链接
So you’re essentially allocating compute to web search in order to save compute on the model.
—— Parag Agrawal · [07:36]
你知道,如果你是 Google,拥有 90% 的搜索市场份额,你是不会再增长的。
指向原始笔记的链接
You know, if you’re Google and you have a 90% search market share, you don’t grow.
—— Cory Doctorow · [13:35]
代码编写智能体其实并不需要你在搜索 Google 时看到的那 10 条蓝色链接。
指向原始笔记的链接
Coding agents don’t really need these 10 blue links that you see when you search Google.
—— Jakub Hojsan · [00:29]
仅仅给你的智能体加一个搜索工具是不够的。
指向原始笔记的链接
Now, adding a search tool to your agent is not enough.
—— Jakub Hojsan · [03:20]
通过为你提供互联网上几乎任何页面的内容,我们给搜索调用增加的额外延迟为零。
指向原始笔记的链接
We add zero extra latency to our search call by providing you the contents of pretty much any page on the internet.
—— Jakub Hojsan · [06:26]
因为说到底,这一切真正有趣的地方在于,如果说我们对 LLM 有任何了解的话,那就是它们太喜欢搜索了。
指向原始笔记的链接
Because at the end of the day, right, what’s really funny about all this is that if we know anything about LLMs, it’s that they just love to search.
—— Dan Adler · [07:23]
成功率是最容易展示的数字,但 Google 的技能其实是在缩小搜索空间。
指向原始笔记的链接
The success rate is the easiest number to show, but Google’s skill is narrowing the search space.
—— Rafal Wilinski · [13:36]
但这里的要点和收获是:搜索帮你找到信息,而 harness 给你可信赖的结果。
指向原始笔记的链接
But the idea here and the takeaway is that search helps you find information, and the harness gives you the results that you can trust.
—— George He · [22:49]
② 出现在这些集
1 集
- 《AI 搜索量今年将超过人类:Exa 的完美搜索宣言》 — 作为概念
③ 关联
点进去有真内容 —— 本页主要出口
Will Bryk · Exa · Cursor · HubSpot · Google · ChatGPT · 智能体 · 嵌入模型 · 搜索 API · LLM
