① 他说过的话

10 条

然后问题就变成,智能体能修复 issue 吗?能,显然能。问题只是,它修得好不好?
And then the question is, can an agent fix issues? Yes, obviously. The question is just, does it fix them well?
—— Mario Zechner · [03:48]

指向原始笔记的链接

任何涉及设计或架构的东西,你还是需要一个人类,因为智能体只会在你的代码库里局部地乱搞一通,把事情弄得更糟。
For anything that concerns the design or architecture, you still kind of need a human because the agent will just locally shit into your code base and make things worse.
—— Mario Zechner · [04:01]

指向原始笔记的链接

而那时智能体通常很擅长找出一堆烂东西,但它非常不擅长提出一个通用的解决方案。
And then usually the agent is great at identifying a bunch of shit, but it’s very bad at coming up with a generalized solution.
—— Mario Zechner · [07:14]

指向原始笔记的链接

地球上没有任何逆向工程师曾经能做到那个程度,尤其是在那么短的时间内。
No reverser on earth would have gone that far ever, especially not in that amount of time.
—— Mario Zechner · [11:06]

指向原始笔记的链接

如果你有一个 oracle 能告诉 LLM 它做的事情是对还是错,那你基本上就赢了。
If you have an oracle that can tell the LLM if what it did is correct or not, then you basically won.
—— Mario Zechner · [13:31]

指向原始笔记的链接

我认为 DHH 不理解机器学习在做什么。
I think DHH doesn’t understand what machine learning does.
—— Mario Zechner · [15:41]

指向原始笔记的链接

那是编程中唯一通常没有以文本形式编码在某个地方、因而 LLM 无法从中学习的东西,也就是构建某个东西架构的过程。
That is the part, that’s the only part in programming that is actually not usually encoded somewhere in textual form so an LLM can learn from it, that the process of building the architecture of something.
—— Mario Zechner · [16:29]

指向原始笔记的链接

也有一种对软件工程未来的愿景,而且他们不确定自己的愿景是否正确,但他们愿意押上全部身家去赌他们的愿景。
Kind of also have a vision of the future of software engineering, and they are unsure of whether their vision is correct, but they’re willing to bet their house on their vision, right?
—— Mario Zechner · [41:23]

指向原始笔记的链接

我认为 Pi 是自我可修改的智能体软件的一个初步概念验证,已经验证通了。
I think Pi is an initial proof of concept for self-modifiable agentic software, worked out.
—— Mario Zechner · [46:50]

指向原始笔记的链接

那是最大的一个问题,却没人去做任何事。
That is the biggest fucking problem, and nobody does anything.
—— Mario Zechner · [55:36]

指向原始笔记的链接

② 出现在这些集

1 集

• 《智能体崩溃后还能原地复活？聊聊 Pi Durable》 — 作为嘉宾

③ 他谈到的

点进去有真内容 —— 本页主要出口

Armin Ronacher · Pi · PyDurable · Arendale · 智能体 · harness · 持久性 · 幂等性 · 效果三明治 · Oracle

④ 也在聊「智能体」的人