集里怎么说它

• 《GPT 5.6 测评：我为什么抛弃 Fable，把 Soul 当主力》(06:46起):本集用来形容 AI 模型生成的千篇一律、毫无必要的内容(如破折号、廉价渐变色、俗套 emoji 等),主持人对此深恶痛绝,并夸奖 Soul 成功避开了这种俗套。
• 《用 AI 对抗 AI：一种不用读代码的编程语言 BAML》(00:13起):本集把它说成:你不读的代码就是 slop;团队干脆废掉代码审查,转而造一堆 sloppy 工具,主动用 slop 对抗 slop。
• 《终结 AI 垃圾内容：Taste Labs 如何度量并对抗 slop》(00:22起):本集的核心敌人：AI 生成内容重复、缺乏契合度、低意图的质感；Thais 认为它比「出色」更容易定义，且被探针证明是可量化的真实现象。

① 提到它的金句

20 条

我认为开源目前更大的问题是，它正某种意义上被善意的泛滥和来自那些试图贡献的人的大量劣质内容扼杀
I think the bigger problem for open source is at the moment, it’s sort of being killed with kindness and a lot of slop from people that are trying to contribute
—— 嘉宾 · [00:08]

指向原始笔记的链接

使用这些工具保持高速度的一个很大部分，就是不允许垃圾代码进入代码库。
A big part of maintaining high velocity with these tools is just not permitting slop to enter the code base.
—— Ryan Lopopolo · [11:36]

指向原始笔记的链接

我通常寻找对这门学科的明显掌控，但还要有品味，能说：“嘿，你要拥有无限的 token，我们不能只是在制造垃圾。”
I generally look for obviously command over the discipline, but then the taste to say, “Hey, you’re going to have unlimited tokens and we can’t just be doing slop.”
—— Andrew Ambrosino · [31:20]

指向原始笔记的链接

在我看来，内容机器真正产出垃圾内容的唯一情况，更多是指那个人在面试环节没分享出足够好的想法，而不是 AI 写出了糟糕的东西。
The only time in my view that the content machine actually produces slop is more of an indictment of the person not sharing good enough ideas during the interview step than the AI writing bad stuff.
—— Alex Lieberman · [00:59]

指向原始笔记的链接

我的看法是，AI 垃圾内容这事的滑稽之处在于，人们其实只是在指着自己说：我不够聪明。
My take is that AI slop is hilariously people just pointing the finger at themselves and saying, I’m not intelligent enough.
—— Alex Lieberman · [01:12]

指向原始笔记的链接

回到之前那个观点，哪怕你相信 AI 会制造垃圾、不让它替你起草，仅仅是让它帮你从空白页走到具体想法，我认为就已经把摩擦降低到你能真正养成创作习惯的程度。
And going back to the point of even if you believe that AI will create AI slop and you don’t want it drafting your content, just helping you go from blank page to concrete idea, I think lowers the friction so much that you’ll actually get in the habit of creating.
—— Alex Lieberman · [15:34]

指向原始笔记的链接

slop(低质内容)的概念对我来说非常有趣,因为我认为它是随着 AI 出现的,但我的看法一直是,有大量人类生产的 slop。
The slop concept is very interesting to me because I think it came out around AI, but my take at least has been there’s a ton of human produced slop.
—— Justine Moore · [22:21]

指向原始笔记的链接

为了击败劣质代码,我们必须成为劣质代码。
To defeat the slop, we must become the slop.
—— Vaibhav Gupta · [01:19]

指向原始笔记的链接

劣质代码就是任何你不阅读的代码。
Slop is just any code you don’t read.
—— Vaibhav Gupta · [01:30]

指向原始笔记的链接

无论你们承不承认,这都是你们的代码库曾经拥有的最少量的劣质代码。
Whether any of you admit it or not, this is the least amount of slop that your code base will ever have.
—— Vaibhav Gupta · [01:32]

指向原始笔记的链接

这就是烘烤进语言里的垃圾,不管你喜欢不喜欢。
This is just slop baked into the language, whether you like it or not.
—— Vaibhav Gupta · [08:14]

指向原始笔记的链接

但你以为你能用这种垃圾糊弄我吗？
But you thought that you could like fool me with this like slop?
—— Wes Bos · [25:47]

指向原始笔记的链接

现在有太多 AI 垃圾内容,你真的需要在组织内有品牌的守护者。
There is so much AI slop out there that you really need custodians of the brand within organizations.
—— Cliff Obrecht · [08:08]

指向原始笔记的链接

所以你可以称之为前 AI 时代的人类垃圾内容。
And so you can call it pre-AI human slop.
—— Parag · [18:42]

指向原始笔记的链接

而且我想说,有些 AI 垃圾内容实际上比我在 AI 出现之前见过的人类垃圾内容要好。
And I would argue that some AI slop is actually better than human slop that I’ve seen before AI.
—— Tyler Folkman · [28:43]

指向原始笔记的链接

如果你不设干预阈值，智能体要么制造垃圾内容，要么烧出极高的 token 账单，因为它只会一直试下去，要么它们会伪造测试，就为了让测试通过。
If you don’t do that, agents will either um create slop or run uh very high token bill because it’s just gonna keep trying, or they’re gonna fake the tests uh just to to to have the test passed.
—— Ran Arusi · [12:37]

指向原始笔记的链接

我觉得有时候很难定义什么是出色，但从大多数人都会同意的意义上说，定义什么是 slop 相当容易。
I think it is hard to define what is great sometimes, but I think it’s pretty easy to define what is slop in the sense that most people would agree.
—— Thais Castello Branco · [03:22]

指向原始笔记的链接

顺便说一句，这比大多数 LLM-as-a-judge 方法——也就是让一个 LLM 来判断那是优质的人类作品还是 AI 生成的 slop——表现都要好。
This performed better, by the way, than most LLM-as-a-judge methods of asking an LLM to judge if that is great human quality versus AI-generated slop.
—— Thais Castello Branco · [08:29]

指向原始笔记的链接

为什么不直接检索一个已经过通盘考虑、具有连贯性的梦幻品牌系统，而不是在那个时刻用生成的方式做出来的东西最终可能不太理想，或者又落入那些 slop 的特征里。
Why not retrieve a dreamy brand system that already has been thought out to be cohesive instead of doing like a generative approach the moment of that might end up not so great or might end up again in those pillars of slop.
—— Thais Castello Branco · [13:14]

指向原始笔记的链接

我们确实不知道如何正确地使用这些工具，而且太容易变成我们所说的「垃圾代码炮」。
We definitely did not know how to properly wield these tools, and it was way too easy to become what we call a slop cannon.
—— Jonathan Kelley · [06:45]

指向原始笔记的链接

② 出现在这些集

3 集

• 《GPT 5.6 测评：我为什么抛弃 Fable，把 Soul 当主力》 — 作为概念(提及)
• 《用 AI 对抗 AI：一种不用读代码的编程语言 BAML》 — 作为概念
• 《终结 AI 垃圾内容：Taste Labs 如何度量并对抗 slop》 — 作为概念

③ 关联

点进去有真内容 —— 本页主要出口

智能体 · Codex · Alex Finn · Vaibhav Gupta · Thais Castello Branco · 系统提示词 · BAML · Taste Labs · gpt 5.6 · 代码审查