cunter1997

就很牛逼

[cunter1997 @cunter1997]

In Shakespeare’s day, most people were surrounded by words. Only one became Shakespeare.

Today most people have books. Few become great writers.

In the coming age of AGI, everyone will have intelligence on tap. What will you do with it?

[cunter1997 @cunter1997]

@neostify 这个可以直接再按折扣的价格再邀请新账号吗？还是只有主号和邀请的账号是这个折扣？如果想要更多只能再弄一个车头？

[cunter1997 @cunter1997]

@MinLiBuilds 你用了语言转文字输入法很可能是 typeless，并且习惯把 claude 读作 cloud

[cunter1997 @cunter1997]

你们的 claude code buddy 是什么？我的是超稀有的隐藏款✌️

[cunter1997 @cunter1997]

many of the remaining challenges may be reducible to smaller subproblems that these methods can solve individually or in combination.

[cunter1997 @cunter1997]

It increasingly feels like most LLM limitations can be traced back to two core bottlenecks: reasoning and context.
Reasoning has been advanced substantially through RL and data scaling. Context limits, meanwhile, are being addressed through tool use, sub-agent decomposition, and

[cunter1997 @cunter1997]

importantly, action–result feedback loops that let the model iteratively act, observe, and refine. If that framing is roughly correct, then the path toward an AGI-like system is becoming easier to see:

[cunter1997 @cunter1997]

@huangyun_122 理论无上限

[Derya Unutmaz, MD @DeryaTR_]

My new Sunday morning routine:

1. Get coffee
2. Check GPT-5.4 projects on the Codex App, continue & start new ones
4. Launch ChatGPT 5.4 Pro for fresh brainstorming sessions
5. Think/learn how to use the 90% of AI capabilities I have yet to explore
6. Drink more coffee

[Penny777 🦊 @penny777]

AI 最大受益群体：爱思考的懒人。

[François Chollet @fchollet]

All the great breakthroughs in science are, at their core, compression. They take a complex mess of observations and say, "it's all just this simple rule".

Symbolic compression, specifically. Because the rule is always symbolic -- usually expressed as mathematical equations. If it isn't symbolic, you haven't really explained the thing. You can observe it but you can't understand it.

[cunter1997 @cunter1997]

I learned a really cool way to use cloud code or the Gemini CLI. CD to the software's user configuration folder, tell CC or CLI to find the software's user configuration file, let it check this file, and then you can adjust any software settings through a conversational interface

[Philo @Philo2022]

Vibe Coding 天然适合 ADHD，传统写代码方式很反 ADHD 个性，Vibe Coding 一下子治好了不爱写测试、抗拒 CSS、老写错变量名、不爱重复琐碎找无聊 bug 的毛病。

现在有 AI 协助，反馈速度变得飞快，成就感满满，越是琐碎重复的部分 AI 越能完成得越好。

然后只用专注于提供大方向、架构设计，持续作为一个项目 leader 存在了。

我并不想亲自敲下每一个字符，但是我很享受创造作品的乐趣。从汇编到高级语言，程序员创造程序的体验应该是越来越好、效率越来越高的。

现在可以以一个纯粹创造者的视角，而不是搬砖工的感觉去创造程序，作为程序员其实应该是非常幸福的。

[Chris Prucha @chrisprucha]

Ok, I'll bite. Programming is NOT dead—it's evolving into something far more powerful than you can imagine.

The next few years will be years of ridiculous individual leverage: one engineer with O3+ will accomplish what once required teams of 50+ engineers.

You will no longer need an organization to create and maintain massive codebases. One engineer will be empowered to create and maintain the most complex software humanity creates—like advanced CAD and simulation software.

Let's be optimistic!

[cunter1997 @cunter1997]

@TheXeophon they will

[cunter1997 @cunter1997]

严正声明：转发不代表本人理解推主的推文含义。另外，强烈谴责此作者，因p图可能造成信息混淆🧐

[王局志安 @wangzhian8848]

大选结果出笼

[cunter1997 @cunter1997]

@WankyuChoi Agree！！

[cunter1997 @cunter1997]

@mutableai People are always worried that AI will encounter challenges when facing a code base with frequent iterations and lagging developer documentation. However, AI solves this problem on its own.

[cunter1997 @cunter1997]

@andrew_n_carr People are always worried that AI will encounter challenges when facing a code base with frequent iterations and lagging developer documentation. However, AI solves this problem on its own.

[cunter1997 @cunter1997]

@_akhaliq I feel this paper will give rise to an independent new field, entirely from the perspective of language, enabling LLMs to understand and generate visual information.