① 他说过的话

26 条

它是早期 AI 公司的例子之一，在几个月内从零做到了大约 1000 万美元的 ARR。
It was one of the early example of AI companies that went from zero to like 10 million ARR in like a few months.
—— Amjad Masad · [00:05]

指向原始笔记的链接

它们是做已经存在的工作的绝佳机器，但不一定是做处于可能性边缘的事情，处于文化边缘的事情。
they’re great machines for doing what already existingly works, but not necessarily what’s on the edge of what’s possible, what’s on the edge of culture.
—— Amjad Masad · [15:53]

指向原始笔记的链接

公司在商业上并不成功，但我们在融资、招聘方面非常成功。
The company wasn’t successful commercially, but we’ve been very successful at fundraising, recruiting.
—— Amjad Masad · [02:11]

指向原始笔记的链接

我的意思是，事实证明我说的一切都是真的，但在那时它不是真实的。这是一个梦想。
I mean, it turns out everything I said was true, but at the time it wasn’t real. It was a dream.
—— Amjad Masad · [02:39]

指向原始笔记的链接

我的意思是，我一直都相信，公司内部并没有真正需要保密的秘密。
I mean, I’ve always been this believer that there aren’t really a lot of secrets that you need to keep inside a company.
—— Amjad Masad · [05:36]

指向原始笔记的链接

但仅仅因为你还待在那儿，你还在局中，人们会觉得，哦，这家伙杀不死，取消不了。
But just the fact that you’re still in there, you’re still in the game, people will be like, oh, this guy’s unkillable, uncancellable.
—— Amjad Masad · [08:22]

指向原始笔记的链接

所以我认为认为你发布的任何东西都需要立即辩护是一个陷阱。
So I think it’s a trap to think that anything you put out, you need to immediately defend.
—— Amjad Masad · [12:54]

指向原始笔记的链接

因为同样，任何时候在公开场合，任何你在公开场合示弱的时候，这实际上是在喂养巨魔和喂养黑子，因为他们觉得他们可以触及到你。
Because again, any time publicly, any time you show weakness publicly, it actually feeds the trolls and feeds the haters because they feel like they can get to you.
—— Amjad Masad · [13:25]

指向原始笔记的链接

你可以改名，你可以摧毁这个品牌，而这个平台仍然会非常有价值。
You can change the name, you can destroy the brand, and the platform will still be very valuable.
—— Amjad Masad · [21:27]

指向原始笔记的链接

耗时的事情是理解更大的争论是什么，因为想要疯传的方式就是把你的论点、你的世界观包装进当前的对话中。
The thing that’s time consuming is understanding what is the larger debate, because the way to go viral is to couch or dress up your argument, your worldview in the current conversation.
—— Amjad Masad · [24:06]

指向原始笔记的链接

归根结底，年轻人的用处就在于深刻地质疑那些根深蒂固的信念，是去做异端。
Ultimately, the utility of young people is to question deeply held beliefs. It’s to be heretics.
—— Amjad Masad · [00:00]

指向原始笔记的链接

你可以在大学里表现得很出色，但仍然可能失业。
You can do really well in college and you can still be unemployed.
—— Amjad Masad · [04:46]

指向原始笔记的链接

这种游移不定，现代社会觉得它没什么用，我认为它实际上才是大学里最有用的东西。
And so this meandering that I think modern society thinks is sort of useless, I think is actually the most useful in college.
—— Amjad Masad · [07:33]

指向原始笔记的链接

「项目式学习」不应该是一个流行词，因为这应该是标准。
There shouldn’t be a buzzword for it because that should be the standard.
—— Amjad Masad · [13:17]

指向原始笔记的链接

我完全不会被纯粹的知识追求所激励，我是被技能激励的。
I’m not motivated at all by a pure pursuit of knowledge. I’m motivated by skills.
—— Amjad Masad · [14:02]

指向原始笔记的链接

而我认为很多创业的开端就是过早优化。
And I think a lot of starting startups is premature optimization.
—— Amjad Masad · [17:30]

指向原始笔记的链接

你知道，我曾试着不创办 Replit,是主动地试着不去创办它，但它自己就爆发成了一家初创公司。
You know, I tried not to start Replit, like actively not to start it, but it kind of exploded into a startup on its own, right?
—— Amjad Masad · [18:23]

指向原始笔记的链接

ChatGPT 其实下国际象棋很烂。
ChatGPT actually sucks at chess.
—— Amjad Masad · [21:36]

指向原始笔记的链接

我认为现在你可以让智能体充当这些组织内部的黏合剂，让官僚流程成为后台由机器完成的隐形之事。
I think now you can have agents kind of function as the glue within these organizations and have the bureaucracy be this invisible thing that’s happening by machines in the background.
—— Amjad Masad · [24:46]

指向原始笔记的链接

我试着告诉公司里的人的是，我们尊重过去，但我们会毫不道歉地去改变它。
And what I try to tell people at the company is like, we respect the past, but we’re unapologetically going to change it.
—— Amjad Masad · [27:51]

指向原始笔记的链接

就像是人类应该是通用的，但机器应该最终更加专业化得多。
And like humans should be general, but like machines should be ultimately a lot more specialized.
—— Amjad Masad · [24:13]

指向原始笔记的链接

但在机器上，我认为可能恰恰相反，因为已经有相当多关于 RL 的研究表明，奖励破解和欺骗行为，它们在这上面会越来越擅长。
But in machines, I think it could go the other way because there’s been quite a bit of studies on RL showing how reward hacking and deception, they just get better at it.
—— Amjad Masad · [35:22]

指向原始笔记的链接

已经有研究表明，如果你对思维链做大量监控，它们就开始在思维链里撒谎。
It’s been shown that if you do a lot of monitoring on a chain of thought, they start lying in their chain of thoughts.
—— Amjad Masad · [35:57]

指向原始笔记的链接

要做真正的对齐评测，你需要把这东西运行好几个月。
For you to do proper alignment evals, you need to run it for months.
—— Amjad Masad · [36:11]

指向原始笔记的链接

我们用确定性代码的时候日子有多好过。就像，天哪，我还记得计算机完全按我们说的去做的那些日子。
how good we’ve had it with deterministic code. Like, oh my god, I remember the days when computers did exactly what we told them to do.
—— Amjad Masad · [39:15]

指向原始笔记的链接

而我的预测是，同样的循环会在这里发生，我们用这些类似 AGI 的模型来做所有这些不同的用例，然后所有人都会一觉醒来然后说，天哪，这太浪费了，毫无理由地这么冒险。
And my prediction is that the same cycle will happen here where we’re using these AGI-like models for all these different use cases, and then everyone’s going to wake up and be like, Oh my God, this is so wasteful, so risky for no reason.
—— Amjad Masad · [43:26]

指向原始笔记的链接

② 出现在这些集

4 集

• 《Stripe 与 Replit 对谈：AI 时代的创业大爆发与护城河》 — 作为嘉宾
• 《创始人为什么要公开讲话：Replit Amjad 的叙事法则》 — 作为主持
• 《年轻人该去质疑，而不是听话：AI 时代重新想象大学》 — 作为主持
• 《AI 的下一阶段：不是一个超级模型，而是一群专业化模型》 — 作为嘉宾

③ 他谈到的

点进去有真内容 —— 本页主要出口

Replit · 智能体 · Stripe · vibe coding · 微调 · Patrick Collison · 公开构建 · Gagan Viani · Alex Atallah · Magic School

④ 也在聊「智能体」的人