当 AI 开始拥有触觉,我们是否应该重新思考,什么才是真正的具身智能?
近日,Meta Reality Labs Research 开源了 SuperDex,一个专注于机器人接触物理仿真的平台。
灵巧手操作的难点,往往隐藏在那些肉眼难以察觉的接触细节中:指尖接触物体的位置、压力的分布、柔软指尖的形变,甚至细微的受力变化,都可能决定一次抓取能否成功。
SuperDex 将“接触”置于物理引擎的核心,通过计算接触表面的三维力分布,为触觉传感器建模提供更精细的数据支持。同时,它还支持刚体、软体、杆、腱与布料等多种物理对象,并提供机器人建模、场景编辑和强化学习接口。
这意味着,我们正在尝试让机器人不只是看见和抓住物体,而是能够在仿真中更加细致地理解自己究竟触碰到了什么,以及物体如何回应它的动作。
但在我看来,这项技术背后隐藏着一个更加值得思考的问题:
具身 AI,是连接世界模型与现实世界物理规律及因果关系的一座桥梁。
如今,许多搭载 AI 的机械手仍然存在一个根本性的局限:感知维度有限,反馈过于机械化。它们可以识别物体、计算压力、执行动作,却很难将真实人手所拥有的丰富触感,完整、连续地转化为 AI 能够理解的数据。
而我们所处的世界是多维的、动态的。它既遵循客观的物理规律,也包含主观的感知体验。
我们对这个世界的理解,从来不只是来自眼睛所看到的画面,或大脑所进行的推理,而是来自身体与环境之间无时无刻不在发生的交互。
因此,我认为,未来具身 AI 的重要发展方向,或许不只是更强大的AI,而是一个更加完整的“身”。
如何帮助 AI 更加真实、生动、细腻地感知和捕捉世间万物,才是世界模型从数字仿真走向现实世界的重要一步。
试想一下,未来的 AI 身体,是否能够拥有像人类一样整体、连续而丰富的感知能力?
感受到一阵微风拂过每一寸肌肤;
感受到一次伸展中,每一根肌肉纤维、肌腱与骨骼之间的相互牵动;
甚至一次钻心的疼痛所对应的复杂生理信号与本能反应。
人类的身体从来不只是大脑执行命令的工具。它本身就是我们认识世界、理解因果、形成直觉的重要组成部分。
或许,未来 AI 的突破,并不只是让机器学会像人一样思考,而是让它真正学会通过身体去认识世界。
而这又引出了一个更深层的问题:
究竟是智能赋予了身体理解世界的能力,还是身体与世界的交互,塑造了智能本身?
🤣 笑死,各大AI 公司内部架构,很真实
@OpenAI 表面使命,向钱看齐
@AnthropicAI 安全团队最强势,部门墙高,老大天天搞邪门理念
@GoogleDeepMind 天天刷榜,组织混乱产品线各自为政,架构混乱人员游动
X Ai 老马独裁
Microsoft 山头林立,报团混战
紫雲 retweeted
🤣 Funny facts about the Major AI companies’ internal structures
@OpenAI: Mission-driven on the surface, money-driven underneath. 💰
@AnthropicAI : Safety team runs the show. Departments are heavily siloed, while leadership keeps pushing some wild philosophies. 🛐
@GoogleDeepMind DeepMind: Chasing benchmarks all day. Organizational chaos everywhere.🗿
Meta: Every product team runs its own kingdom. Employees keep switching sides. 🏰
xAI: Elon runs the entire show. Period.👑
Microsoft: Corporate Game of Thrones. Everyone’s fighting their own battles. ⚔️
Which one is the most accurate? 😂
紫雲 retweeted
According to the research: AI can solve complex problems in seconds, yet fail to cook an egg.
A recent study on GPT-6-Astra's embodied intelligence revealed something fascinating.
In one experiment, an AI-powered robot correctly recognized that an egg was cooked. But while it was busy thinking about turning off the stove, the egg kept cooking until it burned.
The AI failed to account for one simple fact:
The physical world doesn't pause while you think.
This is where World Models become particularly interesting. Embodied AI is giving intelligence a physical presence. World Models could help it develop a better understanding of physical consequences.
But:
If intelligence is the ability to reason, but survival depends on acting at the right moment, is thinking more always the answer?
Instinct, it is an essential yet often overlooked part of intelligence. It’s not just about knowing how to think, but knowing when to stop thinking and start acting.
Can instinct truly be reduced to algorithms and code?
That’s the deeper question we have yet to answer.
Research: arxiv.org/abs/2610.22471
人脑的大小不过半个篮球,却凝聚了推动整个人类文明前进的智慧。那么你知道如今不断刷新人类曾经创造的纪录、推动人类文明实现又一次飞跃的人工智能,背后承载的是多么庞大的系统么?
视频中展示的,正是这场 AI 革命背后鲜为人知的另一面:
从 @nvidia 下一代 Vera Rubin NVL72 机柜级计算架构,到由数十乃至数百个 GPU 协同运作的高性能计算集群,再到 @Meta、@Microsoft 等科技巨头正在大规模建设的 AI 数据中心,人工智能的每一次进步,都离不开背后庞大的基础设施。
这些系统不仅需要数以万计的高性能芯片,还依赖高速互联网络、复杂的散热系统、持续稳定的电力供应,以及将无数计算单元整合为一个整体的工程能力。
我们所见到的是 AI 在几秒钟内完成推理、生成图像、编写代码,这短短几秒的背后,是一整套规模惊人的工业体系在协同运转。
然而,人脑不过三磅重,这种近乎无止境的规模扩张,究竟何时才能实现从量变到质变的跨越,超越算力的堆砌,诞生真正意义上的智能?
喜欢我的内容的请持续关注,谢谢
紫雲 retweeted
Apple is collapsing from within, no single person can shake the whole company unless it’s Steve Jobs
Replying to @johnternus
@johnternus 正在将苹果@Apple 变成一家失去产品原则的公司
从 iPhone Duo 在发布会采访上高高在上地对同行的贬低 “kind of cheap and crappy”到如今各种大跌眼镜的实机表现,再到Macbook哗众取宠的发光logo以及触摸屏却违背了苹果产品使用设计内核的初衷。
到底是他一人的决策还是苹果早就不再是当初的苹果
Replying to @johnternus
@johnternus 正在将苹果@Apple 变成一家失去产品原则的公司
从 iPhone Duo 在发布会采访上高高在上地对同行的贬低 “kind of cheap and crappy”到如今各种大跌眼镜的实机表现,再到Macbook哗众取宠的发光logo以及触摸屏却违背了苹果产品使用设计内核的初衷。
到底是他一人的决策还是苹果早就不再是当初的苹果
Giving AI glasses a gesture-control ring feels like finally giving a trackpad-only laptop a mouse.
And then—wait. Full color, too?
My rational mind: Wait for the reviews.
Also my mind: Full color. Gesture control. What are we waiting for?
AI could get a lot more playful with @RokidColor
#RokidColor
紫雲 retweeted
What if you could interact with the world around you through AR and AI?
AI glasses are nothing new nowadays. They can hear us , talk back, and answer questions. But where's the fun in that ?
The idea never really intrigued me until @Rokid Color moved the entire AI interface right before your eyeballs , in full, lively color.
The true wonder of smart-glasses has never been how smart they can get:
It's about how they spark our curiosity to interact with the world around us, bridging the gap between us and a fully immersive experience of the real world, with a bit taste of super-intelligence ;) .
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
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。
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
紫雲 retweeted
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。
紫雲 retweeted
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…































