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
Photo by 紫雲 (@Deefui)Photo by 紫雲 (@Deefui)Photo by 紫雲 (@Deefui)Photo by 紫雲 (@Deefui)Photo by 紫雲 (@Deefui)Photo by 紫雲 (@Deefui)Photo by 紫雲 (@Deefui)Photo by 紫雲 (@Deefui)Photo by 紫雲 (@Deefui)
紫雲 retweeted
Why Is No One Afraid of AI World Models Replacing Game Engines? The past two days have been pretty interesting for both Silicon Valley and Wall Street. First, @GoogleDeepMind showcased Genie 4, a world model capable of real-time keyboard controls and synchronized visuals and audio. Then, just three hours before its earnings report, Google rushed to announce Gemini 4 Carbon, claiming it had topped the benchmark rankings. When Alphabet finally released its Q3 earnings, the stock plunged more than 10% in after-hours trading. But here's what's even more interesting. Back in January, @Google 's Project Genie announcement sent @unity shares crashing 24%, as investors feared AI-generated worlds could make traditional game engines obsolete. This time? Unity's stock barely reacted. So what changed?
🤖 Made with AI
4
5
4
6
248
紫雲 retweeted
今日科普:@Groq -快到英伟达@NVIDIA都要拿技术授权的芯片 Groq 的 LPU 路线,核心是用高速片上存储和精确的执行调度,减少数据搬运与等待,让模型更快地生成 token。技术介绍 1000 tok/s的时代会到来么?到时候的瓶颈会转移到哪里
8
6
2
15
644
紫雲 retweeted
看到人们为在Macbook上一枚发光苹果标志的回归欢呼雀跃,甚至比其技术与功能上的实质进步更令他们兴奋,我感到有些悲哀。 @Apple拥有堪称艺术品的设计。但它的艺术性,从来不只在于独具一格的外观,更在于设计背后推动时代的灵感,以及挑战成见、坚持自己判断的那份偏执。那些美好的设计之所以令人难忘,是因为它们曾与新的可能一同到来。 发光的苹果标志是否值得值得怀念,答案是当然。只是,当怀旧成了最令人兴奋的卖点,我禁不住在想:我们对苹果的期待,什么时候从“它将带来什么”,变成了“它能还给我们什么”? 我期待那枚标志重新亮起,也期待它背后的创造力,再次让人眼前一亮。
7
5
2
13
4,508
紫雲 retweeted
Replying to @johnternus
@johnternus brings the led logo back to Macbook, when does @Apple plan to put it on a mobile device? An iPhone with led Logo perhaps
3
4
2
6
2,374
MacBook Pro display with touch after 2 days 😭
444
1,547
309
20,350
622,403
最近のAIは狂ってる。AIに物理世界を学ばせる基盤をAIが作ってくれた(しかもブラウザで動く)。仮想世界でAIエージェント同士が競いながら、物理世界を学んでいく。AIがAIを賢くするループ。彼らが機械の身体を手に入れたら確かにゾッとする。
330
67
2,591
761,776
Unfortunately, thinking outside of the box has never been a strength possessed by Artificial Intelligence yet. >0% is still hopeful, but I can only imagine AI being creative in the next 5 years
AI 真的能像科学家一样,自主发现未知的物理规律吗? 最新研究 LongChang(龙场)构建了 14 个物理规律被改写的模拟世界,让 12 个大模型自主观察、提出假设并设计实验。 📊 结果令人意外: GPT-6-Astra:规律发现率 16.7% 其余 11 个模型:全部 0% 增加实验预算也并未改善发现率 一直以来,无法跳出框架思考仍然是 AI 饱受诟病的问题之一 AI 距离真正的自主科学发现,还有很长的路要走。 📄 Paper: arxiv.org/abs/2610.37649 🔗 GitHub: github.com/EnVision-Research…
3
3
1
12
1,467
Here is the original post teapawt.org/OpenAI/status/21147807…
Replying to @OpenAI
GPT-6.1 Terra substantially outperforms GPT-5.6 Terra on Agents’ Last Exam, a test of how well agents use software at work. It even matches GPT-6 Astra at one-tenth the price per token and finishes tasks over 3× faster on average than both Astra and GPT-6.1 Sol.
2
4
89
GPT-6.1 Terra was just introduced by @OpenAI, delivering superior performance at a significantly lower cost. As AI developers continue advancing their models, a couple of things remain unchanged: 1. Diminishing returns: Higher spending doesn't guarantee proportional performance gains. GPT-6 Astra costs significantly more while delivering comparable results. 2. Top performance comes at a premium: Claude Opus 5.5 leads with approximately 63%, but at nearly $9, its cost-effectiveness remains questionable.
3
4
2
9
573
紫雲 retweeted
AI 真的能像科学家一样,自主发现未知的物理规律吗? 最新研究 LongChang(龙场)构建了 14 个物理规律被改写的模拟世界,让 12 个大模型自主观察、提出假设并设计实验。 📊 结果令人意外: GPT-6-Astra:规律发现率 16.7% 其余 11 个模型:全部 0% 增加实验预算也并未改善发现率 一直以来,无法跳出框架思考仍然是 AI 饱受诟病的问题之一 AI 距离真正的自主科学发现,还有很长的路要走。 📄 Paper: arxiv.org/abs/2610.37649 🔗 GitHub: github.com/EnVision-Research…
31
20
3
124
19,135
With @OpenAI’s GPT-6.1 Terra now available, I asked ChatGPT how it would describe the difference between Astra 6, Sol 6.1, and Terra 6.1. My Takeaways: 🧠 Astra: the deep thinker 🛠️ Sol: the everyday all-rounder 🎯 Terra: the practical executor Which has been your go-to?
GPT-6.1 Terra: delivering fast and capable computer use at a lower price. GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it.
3
5
1
10
435
I heard that skywalker keeps popping out on iPhone Duo
1
2
2
8
619
NVIDIA and Meta are expanding their partnership to improve AI model training and inference performance。 This is particularly interesting given Facebook’s history of being caught in major regulatory controversies, especially over data privacy and platform accountability. As AI agents become increasingly autonomous, security, oversight, and accountability are no longer optional. It seems Meta is taking those lessons seriously as it moves further into the AI era.
NVIDIA and @Meta will expand our partnership on @Muse models to improve inference and training performance. Together, we’re exploring NVIDIA OpenShell for safe AI agent deployment and NVIDIA Sentry for monitoring agent behavior. Meta plans broader deployments of NVIDIA Vera Rubin and Vera CPUs. Read the press release: nvda.ws/4zVYmDY
2
2
7
314
X ALGORITHM UPDATE 🚨 X just stopped cutting new posts first, and that should help small accounts (I check it daily with Claude and share updates worth sharing with you) tl;dr: - X can only check 2,800 posts each time you open your feed - new posts used to be last in line, so they got cut first - now everything gets cut equally, new posts included - posts labeled low quality or misinfo now show a warning, and people can't like, reply or share them - no score changes this time lesson learned: your new posts now get a fair chance to be seen. Your first 2 hours still matter most. full details: - How it worked until this week Every time someone opens For You, X collects around 3,700 posts it could show them: posts from people they follow, popular posts, posts liked by similar people, and brand new posts. But X only has time to check 2,800 of them. So it just took the first 2,800 in line. Brand new posts were at the back of the line. When there were too many, they were the ones left out... and a post that never gets checked can't be shown. - What changed Now everything gets cut by the same amount. 25% too many? Every type of post loses 25%, new posts included. Nobody goes first. - Why small accounts should care Those brand new posts are under 2 hours old, with under 8 likes and under 500 views. That's where most posts from smaller accounts are. My read: they now get a fair chance instead of being cut first. How often X actually hits the 2,800 limit isn't in the code, so I can't tell you how big the difference is. - Low quality and misinfo posts now show a warning If X labels your post as low quality or misinformation, people who don't follow you see a warning on it, and they can't like, reply, repost, bookmark or share it. The code doesn't say what gets a post that label. Here's a recap of what you should be doing: - a new post gets its extra push only in its first 2 hours and first 200 views, so post when your people are online - likes carry more weight than ever at 0.66 each - people sticking around pays, the time-spent score is 50% higher than two weeks ago - a read of 10+ seconds after someone opens your post adds 0.4 - the boost favors the post with the best like rate from its first viewers - accounts up to 50K can get one fresh original lifted to roughly 16th - nothing beats a copy link share, it scores 20 - replies, quotes and DM shares each score 5 - talk with your mutuals, replies between you score 20 instead of 5 - originals are the only posts that travel outside your follower base - the 1st and the 32nd like each open a new route to strangers - your 2nd post in the same feed is worth 62% of the first, so pick your best one - videos still need 10+ seconds to qualify for the video lists - earning a follow adds 4 - a pinned post with a bad link can flag the account for 7 days, worth a quick check What you should NOT be doing: - posts that get labeled low quality or misinfo now carry a warning, and nobody outside your followers can like or share them - video alone won't carry you, an open counts 0.02 and the watch 0 - hide-worthy posts cost more, "not interested" sits at -52.27 - a post at 3am burns its 2 hour window before anyone sees it - a tap that turns into a bounce is worth only 0.3 - replies and reposts don't get past your followers - insults get your post hidden from anyone who doesn't follow you - adult media in 3 of your last 5 posts puts a label on your account - disguised content in videos gets checked from 64 likes, harder at 512 - self-reply chains under big accounts risk being zeroed - one account label and non-followers stop seeing you entirely - low-rated links don't belong in your pinned post - recycled wording can get tagged as copypasta spam - strangers tagged plus a low quality link is a pattern X flags - engagement bait has no size exemption - several posts in a row end up competing with each other - hashtag stacks add nothing - bookmarks are still worth 0 - 48 hours in, For You is finished with a post
620
1,294
996
5,227
2,877,154
Grok Voice 3.0 — reportedly resolving 70% of Starlink support calls on its own. I know I've been talking shit about Grok, but I have to admit, this model actually comes in handy when it comes to handling phone calls, something I'm not particularly good at myself. So I decided to simulate a customer complaint call. I started off by cussing it out, and surprisingly, it handled my attitude pretty well, responding promptly and appropriately without losing its composure. According to the chart, it also costs less than other frontier models while delivering even better results. I'm a pretty objective person when it comes to actual utility, and credit where credit is due — Grok deserves some recognition for this one.
2
3
1
4
673
Today I met up with a friend for a round of golf. Between swings, one question kept lingering in my mind: in an age of rapidly advancing AI, what is there that can never be replaced? At first, I thought about careers, skills, and the space humans will have to make a living in the society of the future. But on second thought, perhaps the question shouldn’t be confined to employment and productivity at all. After turning it over, I thought of sports. The meaning of sports has never been only about speed, height, strength, or winning and losing. If thought is an extension of the human spiritual world, then sports are the art of writing our existence with the body. It is an exploration of the body’s limits, a clash between will and instinct, and a portrait of the human refusal to be bound by innate conditions and natural law, always challenging and surpassing ourselves. This is exactly what moves me most about the Olympic spirit. One day, AI may well be able to run faster and jump higher than humans, and execute every movement with precision far beyond our own. But the value of sports has never been about producing an absolutely perfect result. It lies in how a living, breathing person pursues excellence within their own limitations. Machines may surpass human records, but they cannot replace the meaning of humans surpassing themselves. Perhaps, in an era where efficiency and capability are constantly being redefined by AI, sports serve as a reminder: The value of human existence lies not only in what we can do, but in why we do it.
1
4
2
10
255
I told ChatGPT: "From whatever you know about me , create an image of completely original magical creature that represents my personality." Meet the "Emberweaver". Feed this prompt to ChatGPT and drop your creature in the comments👇
🤖 Made with AI
3
2
1
5
223
Giving full credit to its fancy features and performance, as praised by people online and even by Elon himself (of course, it’s his product), if you’re developing a completely separate and superior AI agent that works better than regular Grok, shouldn’t you at least give people who aren’t convinced by your product a chance to try it first? “Pay, then try”—that’s the first I’ve heard of it. What a laughingstock of a business model. @elonmusk
2
2
3
777
Fine, Apple, you’ve proved to the world that you can manufacture a foldable phone. But before you advertise more next-gen hardware, can we focus a little more on Siri AI? People are using your devices to praise other AI agents. Come on!
1
3
2
6
1,221
Today is iPhone Duo launch day. I woke up at 4 a.m. just to witness the launch of this newborn foldable iPhone. Crowds were already forming four hours before the store even opened. When it comes to building that kind of customer loyalty, Apple is one of a kind. #iPhoneDuo
4
2
1
6
606