紫雲 (@Deefui) header
紫雲 (@Deefui) profile picture

全景AI观察者 / Agent实践派 /AI哲学探索者 MBA硕士 资深投资者 AI创业者

Joined May 2025
Pinned Tweet
普罗米修斯把火种带给人类,让我们得以照亮黑夜、锻造工具,创造文明。 AI 时代,我们被赋予了新的火种,能力的边界正在被重新熔铸。 当执行的门槛不断降低,真正限制我们的,不再只是能做什么,而是能想到什么。 我叫紫雲,会在这里记录对 AI 的观察、实践与思考,一起看看,被赋予AI的想象力,究竟能够释放多大的潜能。
1
3
15
1,566
这是GTA IV实机画面?不,你看错了! 这其实是FSD V15 全新的可视化画面。系统会持续实时呈现车身上方某个虚拟机位所看到的场景。 特斯拉的车载 AI 计算机会重新渲染现实场景。需要转弯时,画面里的空中会出现一排指示方向的箭头,就像玩电子游戏里的任务指引一样。你可以在屏幕上操控你的车辆行驶,第三人称视角,就跟GTA一样。 但区别在于,你只有一条命玩这个🤣🤣
8
5
1
13
4,700
分享一个越过X上AI 检测的小技巧: 1. 随机让AI生成一张图 2.找任意一个免费的视频编辑器 3.随便剪裁一下再保存 4.新保存的图片发在X上就不会被检测到啦 #ANT #Riemann
72
🚨 破大防了,家人们!黎曼猜想被 Claude 攻破了?! 困扰人类167年的数学世纪难题,可能真的要被 AI 拿下了! Anthropic 数学研究员宣布,他与 Claude 合作找到了黎曼猜想的证明,目前仍有待独立验证。 但你知道吗?这道数学难题的意义,远不止数学本身。 它甚至与保护我们数字世界的密码学有着深刻联系: 黎曼猜想研究的是质数分布背后的规律。而质数,恰恰也是 RSA 等现代加密技术的重要数学基础。 这一突破还可能催生新的数学方法,推动密码分析与算法优化,甚至启发全新的加密技术。 但是‼️‼️如果 AI 能持续发现人类尚未掌握的数学方法,未来就可能找到针对现有密码系统的全新攻击方式。 而让我担心的是,这些能力一旦被别有用心的人利用,AI 就可能从攻克数学难题的助手,变成突破网络安全防线的工具。
145
紫雲 retweeted
“Inspired by GTA” Now Tesla is rolling out its cutting-edge FSD V15. This feature gives you a so-called “God’s-eye view.” Don’t you think driving in first person is too exhausting? There are so many angles to cover that you’ve gotta keep shifting your gaze back and forth to make sure you’re driving safely. Now, with V15, you can finally drive in third person! You can even maneuver your vehicle on the screen like you do in GTA. Just be careful—you’ve only got one life in this one lol.
9
8
3
54
13,838
紫雲 retweeted
这是GTA IV实机画面?不,你看错了! 这其实是FSD V15 全新的可视化画面。系统会持续实时呈现车身上方某个虚拟机位所看到的场景。 特斯拉的车载 AI 计算机会重新渲染现实场景。需要转弯时,画面里的空中会出现一排指示方向的箭头,就像玩电子游戏里的任务指引一样。你可以在屏幕上操控你的车辆行驶,第三人称视角,就跟GTA一样。 但区别在于,你只有一条命玩这个🤣🤣
8
5
1
13
4,700
紫雲 retweeted
My mathematician friend is freaking out…… When people hear that Claude might have proved the Riemann Hypothesis, their first instinct is to look for holes in the proof, check the code, and see whether the formal verification actually holds up. But my friend? He couldn’t care less whether the proof is right or wrong. He went straight philosophical . He thinks, that mathematical objects don’t exist independently of human understanding. Theorems aren’t some eternal truths floating out there in the universe. They’re concepts we construct in our minds, shaped by the societies and relationships we live in. So if a proof runs to tens of thousands of pages, and not a single human being can fully understand it, then what exactly have we accomplished? In his view, a proof that nobody can comprehend has no social meaning. And without social meaning, it has no mathematical meaning either. At that point, accepting the proof would basically mean taking an AI oracle at its word and trusting that the “high priest” presenting its revelations isn’t pulling a fast one. While everyone else is still arguing over whether Claude actually proved the Riemann Hypothesis, my friend has already moved on to a completely different level: Even if it did, so what? And honestly, he might be onto something. As AI keeps getting better at mathematics, we may eventually reach a point where machines can prove things that no human being can truly understand. Think about that for a second. We could end up with an entire body of mathematical knowledge that humanity can verify but can no longer comprehend. And when that happens, whose mathematics are we even talking about? Because it sure as hell isn’t mine.
10
6
2
18
3,748
紫雲 retweeted
🚨 破大防了,家人们!黎曼猜想被 Claude 攻破了?! 困扰人类167年的数学世纪难题,可能真的要被 AI 拿下了! Anthropic 数学研究员宣布,他与 Claude 合作找到了黎曼猜想的证明,目前仍有待独立验证。 但你知道吗?这道数学难题的意义,远不止数学本身。 它甚至与保护我们数字世界的密码学有着深刻联系: 黎曼猜想研究的是质数分布背后的规律。而质数,恰恰也是 RSA 等现代加密技术的重要数学基础。 这一突破还可能催生新的数学方法,推动密码分析与算法优化,甚至启发全新的加密技术。 但是‼️‼️如果 AI 能持续发现人类尚未掌握的数学方法,未来就可能找到针对现有密码系统的全新攻击方式。 而让我担心的是,这些能力一旦被别有用心的人利用,AI 就可能从攻克数学难题的助手,变成突破网络安全防线的工具。
🚨 BREAKING: Claude just cracked a 167-year-old mathematical mystery? An Anthropic researcher claims to have solved the Riemann Hypothesis, one of mathematics’ greatest unsolved problems, with the help of Claude. This problem concerns the hidden patterns behind prime numbers, the building blocks of mathematics and a foundation of modern cryptography. If verified, this wouldn’t just be a historic mathematical breakthrough. It could mark a turning point where AI begins discovering knowledge beyond what humans have managed to uncover. And the timing couldn’t be more ironic: Anthropic’s stock is also plunging nearly 10% today , alongside a broader AI selloff. Could this be a potential bomb dropped by Anthropic trying to fire up the market and fight the Wall Street ? #ANT #RiemannHypothesis
4
5
1
6
1,292
🚨 破大防了,家人们!黎曼猜想被 Claude 攻破了?! 困扰人类167年的数学世纪难题,可能真的要被 AI 拿下了! Anthropic 数学研究员宣布,他与 Claude 合作找到了黎曼猜想的证明,目前仍有待独立验证。 但你知道吗?这道数学难题的意义,远不止数学本身。 它甚至与保护我们数字世界的密码学有着深刻联系: 黎曼猜想研究的是质数分布背后的规律。而质数,恰恰也是 RSA 等现代加密技术的重要数学基础。 这一突破还可能催生新的数学方法,推动密码分析与算法优化,甚至启发全新的加密技术。 但是‼️‼️如果 AI 能持续发现人类尚未掌握的数学方法,未来就可能找到针对现有密码系统的全新攻击方式。 而让我担心的是,这些能力一旦被别有用心的人利用,AI 就可能从攻克数学难题的助手,变成突破网络安全防线的工具。
🚨 BREAKING: Claude just cracked a 167-year-old mathematical mystery? An Anthropic researcher claims to have solved the Riemann Hypothesis, one of mathematics’ greatest unsolved problems, with the help of Claude. This problem concerns the hidden patterns behind prime numbers, the building blocks of mathematics and a foundation of modern cryptography. If verified, this wouldn’t just be a historic mathematical breakthrough. It could mark a turning point where AI begins discovering knowledge beyond what humans have managed to uncover. And the timing couldn’t be more ironic: Anthropic’s stock is also plunging nearly 10% today , alongside a broader AI selloff. Could this be a potential bomb dropped by Anthropic trying to fire up the market and fight the Wall Street ? #ANT #RiemannHypothesis
4
5
1
6
1,292
紫雲 retweeted
分享一个越过X上AI 检测的小技巧: 1. 随机让AI生成一张图 2.找任意一个免费的视频编辑器 3.随便剪裁一下再保存 4.新保存的图片发在X上就不会被检测到啦 #ANT #Riemann
4
5
1
6
572
分享一个越过X上AI 检测的小技巧: 1. 随机让AI生成一张图 2.找任意一个免费的视频编辑器 3.随便剪裁一下再保存 4.新保存的图片发在X上就不会被检测到啦 #ANT #Riemann
4
5
1
6
572
紫雲 retweeted
🚨 BREAKING: Claude just cracked a 167-year-old mathematical mystery? An Anthropic researcher claims to have solved the Riemann Hypothesis, one of mathematics’ greatest unsolved problems, with the help of Claude. This problem concerns the hidden patterns behind prime numbers, the building blocks of mathematics and a foundation of modern cryptography. If verified, this wouldn’t just be a historic mathematical breakthrough. It could mark a turning point where AI begins discovering knowledge beyond what humans have managed to uncover. And the timing couldn’t be more ironic: Anthropic’s stock is also plunging nearly 10% today , alongside a broader AI selloff. Could this be a potential bomb dropped by Anthropic trying to fire up the market and fight the Wall Street ? #ANT #RiemannHypothesis
12
6
2
10
7,716
紫雲 retweeted
I just discovered this simple trick of getting rid of “AI generated” tag on X. I found no one has shared this on X so I think this might help many of you
6
5
1
6
902
谁把 Monid 的 launch 排在 GTA 6 解锁前夜的?我手柄都拿起来了! 刷到 IGN 给了 10/10,往下一滑,GameSpot 也是 10/10。我第一反应居然是:R 星的股票不得涨疯了?等一下,R 星有股票代码吗? 我就想查个公司背景,总不能先交一年年费吧。于是我打开 Claude,让它用 Monid 查,按请求付费,一次 也就是说10 美分。 原来R 星并没有上市,是 Take-Two 的全资子公司,股票代码 TTWO 是母公司的。 行了,股票代码记下了。今晚先给他们贡献一份销量。🎮 #MonidKilledPitchBook #GTA6
10
12
1
27
7,116
My call for Anthropic’s Nasdaq debut is $90–$100 at the open and around $92 at the close. Long term, I’m more bullish. I think we’re underestimating what Anthropic could eventually sell. For day one, I expect demand to pull the stock above its $83 IPO price. Investors bid the deal $5 above the top of its original range. But up to 315 million shares held by employees and other existing holders are also eligible for sale on day one. If the stock opens at $100, some of those holders have a reason to sell. A $92 close would give someone who got the IPO price an 11% gain and someone who bought at $100 an 8% loss. Both could walk away with a very different opinion of the same IPO. What makes me bullish over the next few years starts with a question: Why stop at selling access to Claude when Anthropic could sell the work Claude produces? Right now, people like us pay a monthly subscription. Businesses also pay to use Claude in their products and workflows. Demand is real. Anthropic generated nearly $17 billion in third-quarter revenue, up 47% from the previous quarter. Yet it still recorded a $567 million operating loss, an operating margin of roughly negative 3.3% across the company. And for every dollar it spent on compute and infrastructure, it generated $1.72 in revenue, down from $2.08 the quarter before. More people using Claude can drive extraordinary growth while leaving shareholders with very little operating profit to show for it. Here’s where I think the business could change. A manufacturer could buy a component design with the test data showing it meets spec. A software company could buy a completed migration that passes its acceptance tests. A biotech firm could license a promising discovery together with the experimental evidence behind it. In each case, the customer buys the finished result as a package. The buyer can put it to work without first subscribing to Claude, hiring a team to use it, and managing the project itself. That’s my long-term thesis going into the IPO. Individual subscriptions helped establish demand for Claude. The development that would make me materially more bullish is seeing companies return to Anthropic to buy a second finished project. That would tell us the first one delivered enough value for the customer to come back.
We’ve priced our IPO at $83 per share, with trading expected to begin on Nasdaq tomorrow. anthropic.com/news/anthropic… Legal disclaimer: A registration statement relating to the Class A common stock was declared effective by the Securities and Exchange Commission on November 18, 2026. The offering is being made only by means of a prospectus. Copies of the final prospectus may be obtained, when available, from: Goldman Sachs & Co. LLC, Attention: Prospectus Department, 200 West Street, New York, NY 10282, by telephone at (866) 471-2526, or by email at prospectus-ny@ny.email.gs.com; or Morgan Stanley & Co. LLC, Attention: Prospectus Department, 180 Varick Street, 2nd Floor, New York, NY 10014, or by email at prospectus@morganstanley.com
2
5
3
7
1,269
紫雲 retweeted
AI时代中,人类游戏工业的最顶级产品 MC均分99分,人类历史上最高分的游戏 前面50年,后面10年都不会出现的游戏 这个游戏我只能用一个词来形容【可怕】【太可怕了】 建模、水体、光影、天气,物理 这个地球上最顶级的水平。 我真的分不清游戏跟现实了…… 无数细节,无数脚本,无数随机事件,一环扣一环的网状叙事 在AI完全侵染游戏开发前,这可能是地球上最后一个手搓3A游戏 玩,必玩,每个人都必须玩!!! #GTA6
1
6
1
13
7,042
AI时代中,人类游戏工业的最顶级产品 MC均分99分,人类历史上最高分的游戏 前面50年,后面10年都不会出现的游戏 这个游戏我只能用一个词来形容【可怕】【太可怕了】 建模、水体、光影、天气,物理 这个地球上最顶级的水平。 我真的分不清游戏跟现实了…… 无数细节,无数脚本,无数随机事件,一环扣一环的网状叙事 在AI完全侵染游戏开发前,这可能是地球上最后一个手搓3A游戏 玩,必玩,每个人都必须玩!!! #GTA6
1
6
1
13
7,042
玩上了兄弟们 #GTA6
1
114
紫雲 retweeted
谁把 Monid 的 launch 排在 GTA 6 解锁前夜的?我手柄都拿起来了! 刷到 IGN 给了 10/10,往下一滑,GameSpot 也是 10/10。我第一反应居然是:R 星的股票不得涨疯了?等一下,R 星有股票代码吗? 我就想查个公司背景,总不能先交一年年费吧。于是我打开 Claude,让它用 Monid 查,按请求付费,一次 也就是说10 美分。 原来R 星并没有上市,是 Take-Two 的全资子公司,股票代码 TTWO 是母公司的。 行了,股票代码记下了。今晚先给他们贡献一份销量。🎮 #MonidKilledPitchBook #GTA6
10
12
1
27
7,116