FRONTIER AI MINDS · ANNUAL BRIEFING
AI 竞争,正在从模型竞赛升级为系统竞赛
六位前沿人物过去一年的公开言论,指向能力、部署与治理的同一场变化。
EXECUTIVE ANSWER
真正稀缺的是把智能转化为结果的完整系统
模型分数只是起点;成本、能源、工作流、执行权限和责任机制共同决定现实领先。
我们想构建所谓的世界模型,即一种真正理解世界物理规律的模型。We want to build what we call a world model, which is a model that actually understands the physics of the world.Demis Hassabis on shipping momentum, better evals and world… · 00:00:14
客户购买的是芯片、网络、电力和冷却共同构成的生产系统。
世界正在竞相建设 AI 工厂;芯片、机架、网络、电源、冷却与电网必须端到端协同设计,因为计算就是收入。
The world is racing to build AI factories. Every layer—chip, rack, network, power, cooling and grid—must be designed together end to end because compute is revenue.NVIDIA GTC Taipei 2026 Keynote · 00:26:52计算量超过某一门槛的模型,应当由具备资质的第三方强制测试。
计算量超过某一门槛的模型,应当由具备资质的第三方强制测试其在四个特定领域的风险水平:网络安全、生物武器、AI 系统失控,以及可能加速上述其他风险的自动化研发。
Models above a threshold of compute should undergo mandatory testing by a qualified third party for their level of risk in four specific areas: cybersecurity, biological weapons, loss of control of AI systems, and automated R&D that could accelerate these other risks.Policy on the AI Exponential · paragraph 15SIX ROUTES · ONE ERA
六个人不是六句口号,而是六套因果链
科学目标、智能供给、安全治理、工业产能、世界模型与递归生产——六条路线在现实部署处相遇。
用更完整的认知系统推进科学与医学。
届时,我们将拥有一种前所未有的工具或系统,必须思考如何使用如此具有变革性的技术。在我看来,它会改变一切:经济,甚至可能包括人类的生存状态。
So around 2030 um and then when that happens you know this that that becomes a new type of tool or uh system that we've never seen before and we'll need to uh uh think about how to use such a transformative piece of technology. uh…The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:18:07从 o1 到 GPT‑5.4,同一难题的推理成本约下降千倍。他据此判断,模型、内核、电力与数据中心仍有巨大的联合优化空间。
人们会引用各种惊人数据,说明我们的模型乃至整个行业的模型随时间变得多么高效;我认为其中一个特别惊人的例子是,我们的第一个推理模型叫 o1,大约在 16 个月前发布。
Um people cite whatever amazing statistic they like about how much more efficient our models have gotten over our industry's models have gotten over time, but one that I think is incredible, our first reasoning model was called O1, came out like 16 months ago.LIVE: Sam Altman speaks at BlackRock's US Infrastructure Su… · 00:09:57高速前进,同时把灾难风险降到接近零。
[笑声] 首先,谁也别再说 Claude 对我手软,或 Claude 偏袒我或 Anthropic。就像,天哪,我当时想 >> Claude,Claude 出的是硬球。
[laughter] First of all, let no one ever say that Claude is soft on me or Claude is biased towards me or towards Anthropic. Like, you know, Wow, I thought >> Claude Claude throws hard balls.The Co-Founders of Claude AI Tell Oprah About the Impact Ar… · 00:02:37把 Token 作为新工业体系的产出。
如果有 1 吉瓦电力,每瓦吞吐量就是收入,因为每个 Token 都有利润;计算就是收入,每瓦性能就是收入。
If you have 1 gigawatt of power, throughput per watt is revenue because every token is profitable. Compute is revenue. Performance per watt is your revenue.NVIDIA GTC Taipei 2026 Keynote · 00:36:49通过世界预测和行动规划走向智能。
智能包括理解世界中发生了什么、世界将如何演变、你的行动可能产生什么后果,从而让你能够规划一系列行动来达成特定目标。而这是大语言模型完全做不到的。
Being intelligent includes understanding what happens in the world, how the world is going to evolve, what the consequences of your actions might be, which would allow you to plan a sequence of actions to arrive at a particular goal. And this is something that LLMs…Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 21数字与物理机器人形成生产扩张。
好吧,如果数字人体仿真问题到今年年底还没有得到解决,我会感到惊讶。我想这就是我们所说的 MacroHard 项目的意思。
Well, I'd be surprised by the end of this year if digital human emulation has not been solved. I guess that's what we sort of mean by the MacroHard project.马斯克如何把 1 太瓦 GPU 送上太空 · 01:00:39READING CONTRACT
统计只占这一页;之后每个结论都回到原话
97 篇双语材料、805 条去重观点、43 个人物—主题正式轴。公开言论能证明“他反复强调什么”,不能直接证明技术已实现。
时间表必须连同“仍可能需要重大突破”的条件阅读。
以及你认为它会带来哪些益处、时间表又如何?我知道你说的是 2030 年及以后,但希望你再展开一些。
will change with AGI uh >> and the benefits as you see it and also the time scales. I know you say 2030 and onwards but a little bit of color on that would be very good.The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:16:50愿景与非零风险可以同时出现,不能只摘一侧。
>> 我若认为自己能控制远比我聪明的超级天才 AI,可能只是虚荣。我们能做、也应努力做的是确保 AI 具有良好价值观,关心人类,并且——
>> I give you the Levi Strap. >> I think I think I would probably be it would be an exercise in vanity to for me to think that I would be controlling super genius AI that is vastly smarter than me.The full-length interview with Elon Musk | The Economist · 00:05:23DEMIS HASSABIS
目标函数是科学,AGI 是实现目标的工具
科学与医学目标 → 补齐认知能力 → 世界模型与 Agent → 安全部署。
AGI 最重要的正向用途指向科学、医学和治病。
一两项重大突破,就像 Transformer,或我们在 AlphaGo 中把强化学习用于深度学习那样。因此,答案尚未揭晓。
or two big breakthroughs still uh like transformers were or things like reinforcement learning applied to deep learning like we did with Alph Go. So the jury's out on that.The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:18:07从蛋白质结构这一根节点打开生物学与药物发现。
——德米斯,我再问最后一个问题:为什么你会选择蛋白质折叠作为首个攻克目标?——我毕生研究 AI,是因为我认为它最终会成为科学的终极工具。
>> And and just one more question for you Dennis. Uh I was very keen to understand why did you choose protein folding >> as the first uh >> Yes.The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:42:38当前系统仍缺持续学习、长期记忆与规划。
你认为在实现它之前仍缺少哪些能力?我认为还缺少几样东西。
So I think there's still a few things missing. I feel like we still have kind of jagged intelligences in the sense that they're amazingly good at certain things.Demis Hassabis on AGI by 2030, Curing Every Disease, Life A… · 00:03:21SAM ALTMAN
智能是一条不断降价并向外扩散的供给曲线
能力扩展 → 推理降本 → 基础设施扩容 → 平台分发 → 社会吸收。
企业应按未来两到四年的能力与成本曲线设计。
>> 我认为,如果启动初创公司的类似障碍和类似的工具变得更加激增,所有这些东西,你会认为像更困难的初创公司更容易启动。
>> I would assume that if the like barriers to starting startups and the like tools out there become more proliferated and all this stuff, you would assume that like harder startups are easier to start.Sam Altman - How to Start a Startup · 00:01:12从 o1 到 GPT‑5.4,同一难题的推理成本约下降千倍。他据此判断,模型、内核、电力与数据中心仍有巨大的联合优化空间。
人们会引用各种惊人数据,说明我们的模型乃至整个行业的模型随时间变得多么高效;我认为其中一个特别惊人的例子是,我们的第一个推理模型叫 o1,大约在 16 个月前发布。
Um people cite whatever amazing statistic they like about how much more efficient our models have gotten over our industry's models have gotten over time, but one that I think is incredible, our first reasoning model was called O1, came out like 16 months ago.LIVE: Sam Altman speaks at BlackRock's US Infrastructure Su… · 00:09:57模型、ChatGPT、API 与 Codex 构成通用平台。
我们的战略是构建核心模型和基本使用方式——ChatGPT、API、Codex 等智能体以及通用智能体——并真正赋能其他所有人进行建设。我从 OpenAI 学到的一件事是,世界上大多数创新都来自初创企业,而我们希望赋能它们。
Um we way more innovation is going to happen outside of OpenAI than inside of OpenAI and our strategy is to build the core model and the basic way to use it, ChatGPT, the API, agents like Codex and the general agent, and really empower everybody…Fireside Chat: Sam Altman×Vinod Khosla and AMA at IIT Delhi… · 00:33:05DARIO AMODEI
安全不是减速口号,而是能力增长假设的后果
能力快速增长 → 科学与企业价值放大 → 就业、滥用与失控风险同步放大 → 测试与责任前置。
Scaling 会让机会与风险同时扩大。
是的,回到 Anthropic 的创立:我和 Anthropic 的联合创始人是最早记录技术规模扩展及其将变得多么强大的人之一。我们创办公司背后的想法是,看着技术进步如此迅速,我们说:“看,我们确实需要把这件事做好。
Yes, so, you know, go going back to the founding of Anthropic, you know, we were among the first myself and my co-founders at Anthropic to to document the scaling of the technology and how powerful it would become. And you know, the idea behind how…「不毛な競争には加わらない」アンソロピック ダリオ・アモデイCEO 単独インタビュー「安全なAI」と日本市場への期待【… · 00:01:32技术列车不能由一家停止,也不能任其失控。
嗯,真正回答这个问题的话,这项技术是人类一个划时代的变化。你知道,某个时候人类开始使用火;某个时候,人类造出了蒸汽机和工厂,由此创造了工业革命。
You know, at some point humans started to, you know, started to use fire. At some point humans, you know, built built steam engines and factories, and that created the Industrial Revolution.The Co-Founders of Claude AI Tell Oprah About the Impact Ar… · 00:02:37高增长可能与高失业、低工资和不平等并存。
我们或许会看到一个极不寻常的组合:GDP 高速增长,同时伴随高失业、至少是就业不足,或大量低工资岗位。
I think we could have this very unusual combination of very fast GDP growth and high unemployment, or at least underemployment or, you know, low wage job.Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:27:41JENSEN HUANG
AI 不是软件功能,而是一种新工业产能
扩展律 → AI 工厂 → Token 产出 → Agent 与物理 AI → 行业生产率。
芯片、机架、网络、供电与冷却联合构成 AI 工厂。
世界正在竞相建设 AI 工厂;芯片、机架、网络、电源、冷却与电网必须端到端协同设计,因为计算就是收入。
The world is racing to build AI factories. Every layer—chip, rack, network, power, cooling and grid—must be designed together end to end because compute is revenue.NVIDIA GTC Taipei 2026 Keynote · 00:26:52固定电力预算必须产生更多 Token 收入。
如果有 1 吉瓦电力,每瓦吞吐量就是收入,因为每个 Token 都有利润;计算就是收入,每瓦性能就是收入。
If you have 1 gigawatt of power, throughput per watt is revenue because every token is profitable. Compute is revenue. Performance per watt is your revenue.NVIDIA GTC Taipei 2026 Keynote · 00:36:49当 Agent 能调用子代理、工具和记录系统,企业平台必须提供身份、权限、隐私路由、网络护栏与完整审计。
【Peter】我很喜欢自己的工具赋能普通人;一位 60 岁父亲安装它后,甚至把酿酒机器通过蓝牙接入 OpenClaw。
And he had like one guy, he told me like he installed it as a 60 year old dad and like they made beer, connected the machine via Bluetooth to open cloth.NVIDIA GTC Keynote 2026 · 01:49:15YANN LECUN
对 LLM 的批评只是起点,JEPA 才是替代路线
连续现实信号 → 抽象表示 → 世界预测 → 行动规划 → 机器人与工业控制。
智能必须预测世界演化和行动后果。
智能包括理解世界中发生了什么、世界将如何演变、你的行动可能产生什么后果,从而让你能够规划一系列行动来达成特定目标。而这是大语言模型完全做不到的。
Being intelligent includes understanding what happens in the world, how the world is going to evolve, what the consequences of your actions might be, which would allow you to plan a sequence of actions to arrive at a particular goal. And this is something that LLMs…Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 21在表示空间保留对行动后果有用的信息。
JEPA 的基本思想是:我们观察的是一种复杂现象,不是离散 token 序列,不是语言,而是来自工业系统的传感器数据、视频,或连续、高维且含噪的信号。你不能用生成式 AI 去理解它。
The basic idea of JEPA is that we have a complex phenomenon that we observe, not a sequence of discrete tokens, not language, but sensor data from industrial systems, videos, or continuous high-dimensional noisy signals. You cannot use generative AI to understand it.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 41生成合理后续并不等于理解真实后续。
当然,你会说:“等等,但我们有能进行视频生成的系统。
Now, of course, you're going to say, "But wait, we have systems that can do video generation.Fireside Chat with Yann LeCun, Executive Chairman of AMI La… · 00:06:10ELON MUSK
机器人是经济叙事中心,能源是扩张硬约束
求真模型 → 数字工作自动化 → 物理机器人 → 递归生产 → 能源与太空算力。
数字工作先受冲击,物理工作等待机器人。
>> 我不希望你惊吓大家,但 AI 多快、多大程度取代工作确实是重要辩论。正如你说,十年后人工智能会超过所有人类智能。
That is natural. >> So I I certainly don't want you to alarm people, but I do think there is a really important debate about how quickly and to what degree AI displace people from jobs because it is as you say if in 10 years…The full-length interview with Elon Musk | The Economist · 00:28:27机器人制造更多机器人,形成递归扩张。
好吧,如果数字人体仿真问题到今年年底还没有得到解决,我会感到惊讶。我想这就是我们所说的 MacroHard 项目的意思。
Well, I'd be surprised by the end of this year if digital human emulation has not been solved. I guess that's what we sort of mean by the MacroHard project.马斯克如何把 1 太瓦 GPU 送上太空 · 01:00:39真正可用的人形机器人尚不存在;若同时获得通用能力与量产能力,它可能成为比手机更大的产品类别。
至于人形机器人,目前没有真正有用的人形机器人,只有一些噱头。
Um there are sort of gimmicks, but there are there are no actually useful humanoid robots.FULL: Elon Musk Makes Shocking Future Predictions At U.S.-S… · 00:03:06CAPABILITY ROUTES
规模仍有效,但“继续堆”不是完整解释
能力继续增长的证据很强;长期记忆、世界理解与可靠行动仍没有被规模单独解释。
扩展已不只是预训练:后训练与推理时计算同样决定能力;系统“思考”越久,检索与反复推演越充分。
为此必须进行大量推理,因此在强化学习中,训练与推理已经融为一体。>> 这非常复杂,也就是所谓的后训练。
And so it tries a whole bunch of different ways and and uh in order to do that, yeah, you've got to do inference. So now training and inference are now integrated in reinforcement learning.NVIDIA: OpenAI, Future of Compute, and the American Dream |… · 00:01:52推理、发现与持续学习曾被视为“能力墙”,却不断被新训练方法和更长上下文突破;当前缺陷不应被视为不可跨越。
过去是“AI 模型不能推理”,最近则是“AI 模型不能做出新发现”。
It was like, "AI models can't reason." And recently, there's this, "AI models can't make new discoveries.A Cheeky Pint with Anthropic CEO Dario Amodei · 00:33:27AGI 仍可能需要一到两个重大突破。
首先,我对 AGI 的定义是:一个经测试后能表现出人类所具备全部认知能力的系统。之所以以此为标准,是因为人类及人脑是我们目前所知——也许是整个宇宙中——通用智能存在的唯一实例。
I mean firstly my definition of AGI is is is you know a system that you can test and exhibits all the cognitive capabilities humans have. The reason that's important is um the human reference point and the human brain reference point is the human brain…The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:16:50语言序列预测缺少世界模型与行动后果预测。
智能包括理解世界中发生了什么、世界将如何演变、你的行动可能产生什么后果,从而让你能够规划一系列行动来达成特定目标。而这是大语言模型完全做不到的。
Being intelligent includes understanding what happens in the world, how the world is going to evolve, what the consequences of your actions might be, which would allow you to plan a sequence of actions to arrive at a particular goal. And this is something that LLMs…Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 21WORLD MODELS
“世界模型”其实是三种不同的赌注
Demis 讲认知机制,Yann 讲学习目标,Jensen 讲工业仿真;同名不等于共识。
我们想构建所谓的世界模型,即一种真正理解世界物理规律的模型。We want to build what we call a world model, which is a model that actually understands the physics of the world.Demis Hassabis on shipping momentum, better evals and world… · 00:00:14
以表示空间预测支持行动规划。
原因在于,其中大部分数据不可预测:既有噪声,也有高度复杂的依赖关系。如果你尝试训练一个系统预测视频接下来会发生什么,它不可能预测每一个细节。
And the reason is that most of that data is unpredictable. There’s noise, and there are highly complex dependencies in it.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 41用数字孪生与合成数据训练物理系统。
因此,智能体系统、机器人系统、物理人工智能、数据是最难的问题。你已经看到我们沿着这个阶梯向上移动。
And so, agentic systems, robotic systems, physical AI, the data is the hardest problem. You've seen us move up this ladder.NVIDIA GTC Taipei 2026 Keynote | Full Replay · 01:38:42CAPABILITY GAP
真正未解决的是长期可靠性
榜单、单轮问答与演示不能替代长期任务完成率、失败可逆性、环境变化下稳定性和监督成本。
持续学习、记忆与长期推理仍是缺口。
你知道,我的联合创始人 Shane 等人曾公开说过大约是 2028 年、2030 年。我们把它视为一项 20 年使命,我认为我们正按此推进。
You know, people like Shane, my co-founder, was on record saying around 2028, 2030 you know, we thought of it as a 20 year mission, and I think we're on track for that. What do you think the capabilities that are missing are still before we…Demis Hassabis on AGI by 2030, Curing Every Disease, Life A… · 00:03:21每个新任务都依赖大量示范,说明系统并未理解行动后果;真正泛化要求机器人能预测自身动作造成的未来。
最终,这会带来家用机器人、自动驾驶汽车等许多应用中的智能系统。
Eventually, that will lead to intelligent systems for things like domestic robots, self-driving cars, and many other applications.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 54真实世界数据昂贵、稀缺且覆盖不了长尾;物理 AI 必须在物理一致的闭环仿真中经历数十亿级训练与评估。
物理 AI 的ChatGPT 时刻即将到来。
The Chat GPT moment for physical AI is nearly here.NVIDIA Live at CES with CEO Jensen Huang · 00:31:01FOUR QUESTIONS
“模型很强”必须拆成四个可验证问题
能不能做?能不能持续做?能不能负担?能不能负责地做?
受控测试中的能力只是第一问。
展望今年,AI 故事和成就会是什么?我每年都说 AI 每年都关键;对前沿科学工作者而言,几乎每年都像十年,今年也不例外,强度很高。
So when you look at the year ahead, what do you think the story of AI will be? What will we achieve?Can Google’s New AI Solve Everything? | Titans and Disruptors · 00:25:15当 Agent 能调用子代理、工具和记录系统,企业平台必须提供身份、权限、隐私路由、网络护栏与完整审计。
【Peter】我很喜欢自己的工具赋能普通人;一位 60 岁父亲安装它后,甚至把酿酒机器通过蓝牙接入 OpenClaw。
And he had like one guy, he told me like he installed it as a 60 year old dad and like they made beer, connected the machine via Bluetooth to open cloth.NVIDIA GTC Keynote 2026 · 01:49:15DEPLOYMENT · 1
真正的部署单位是有责任人的工作流
模型价值不会自然释放;目标、数据、工具、权限和责任人必须进入同一个组织闭环。
产品战略从最强基础模型和研究开始,再快速进入产品、营销和分发;Google 正重构基础设施,把最新模型接入核心产品。
拥有最好的模型,由基础研究供给;随后尽快反映到产品,做好营销和分发。但模型若非同类最佳、最先进,其他都不重要。
having the best models in this case, and obviously fundamental research feeding into that, and I always believe you then need to reflect that obviously as quickly as possible in your products, and then you've got to get your marketing and distribution right. But none of…Can Google’s New AI Solve Everything? | Titans and Disruptors · 00:19:37更聪明的模型要进入企业并推进疾病治疗;企业本身是超越个人的集体智能,人机组织的上限可能高于单个模型。
……希望公司最终抵达的位置。
where we hope the the company will be.A Conversation with Dario Amodei and Marc Benioff | Dreamfo… · 00:31:19当 Agent 能调用子代理、工具和记录系统,企业平台必须提供身份、权限、隐私路由、网络护栏与完整审计。
【Peter】我很喜欢自己的工具赋能普通人;一位 60 岁父亲安装它后,甚至把酿酒机器通过蓝牙接入 OpenClaw。
And he had like one guy, he told me like he installed it as a 60 year old dad and like they made beer, connected the machine via Bluetooth to open cloth.NVIDIA GTC Keynote 2026 · 01:49:15DEPLOYMENT · 2
从总算力转向单位有用任务的全成本
推理、能源、网络、人工复核、失败重做与基础设施利用率都要计入。
固定电力要转化为更多可售产出。
如果有 1 吉瓦电力,每瓦吞吐量就是收入,因为每个 Token 都有利润;计算就是收入,每瓦性能就是收入。
If you have 1 gigawatt of power, throughput per watt is revenue because every token is profitable. Compute is revenue. Performance per watt is your revenue.NVIDIA GTC Taipei 2026 Keynote · 00:36:49晶体管、电力、数据中心和机器人共同约束智能供给。
>> 晶体管,其次是电力——按这个顺序。
>> Transistors and then electrons in that order.Sam Altman - How to Start a Startup · 00:10:56xAI 的 50→500MW 设施计划依赖国家、资本、能源与芯片供应商协同;前沿竞争已超出单一实验室可以独立承担的尺度。
埃隆,先谈 xAI。马斯克:我们很高兴宣布,我们在做一个 500 兆瓦——对,500 兆瓦。
Let's begin with you Elon. the things that we're doing with XAI.FULL: Elon Musk Makes Shocking Future Predictions At U.S.-S… · 00:15:40DEPLOYMENT · 3
Agent 的门槛不是会回答,而是能被授权
成熟度要看任务完成率,也要看授权范围、错误半径、回滚能力、监督时间与总成本。
模型能力已接近可委托完整任务的门槛;一旦从建议转向代办,偏离、暂停和恢复就成为产品能力的一部分。
展望今年,AI 故事和成就会是什么?我每年都说 AI 每年都关键;对前沿科学工作者而言,几乎每年都像十年,今年也不例外,强度很高。
So when you look at the year ahead, what do you think the story of AI will be? What will we achieve?Can Google’s New AI Solve Everything? | Titans and Disruptors · 00:25:15部分团队约九成代码由 AI 生成,但人仍负责编辑、监督和最困难部分;自动化比例越高,监督质量越关键。
谈论未来 AI 时,整件事难免带上一层科幻光泽。它作为比喻有帮助,却也可能让人分心,因为它会把所有人的思维带入一种未来幻想模式。
There there's I think inevitably as we, you know, talk about AI in the future, this kind of sci-fi sheen on on on, you know, the on the whole thing. And that's, you know, in a way it's helpful as a metaphor, but also it can…A Conversation with Dario Amodei and Marc Benioff | Dreamfo… · 00:16:03当前 Agent 多是在盲目套用昂贵、难泛化的步骤配方;只有世界模型进入系统,跨任务泛化才可能发生质变。
不,我认为几年内它会变得非常不同,主要是因为当前智能体系统缺乏世界模型。构建一个无法预测自身行动后果的智能体系统,是非常糟糕的主意。
No, I think it’s going to look very different within a few years. Mostly because of the lack of world models in current agentic systems.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 73PHYSICAL AI
机器人是所有技术路线的现实压力测试
世界理解、长期规划、数据、硬件、可靠性、安全和单位成本被同时拉到一张考卷上。
机器人和自动化供应链的战略价值,是让数据中心建设逐步自我扩张,把智能供给的物理瓶颈纳入系统设计。
显然,我们不会真的把一切都放进 OpenAI,但会努力建立一条比迄今更有效运转的供应链。
We're not literally going to try to do it all inside of OpenAI obviously but we will try to do a better job of a wellSam Altman - How to Start a Startup · 00:48:17机器人检验模型能否预测行动后果并泛化。
世界模型的成功对国防这样的行业意味着什么?
What does success at world models mean for an industry like defense?Fireside Chat with Yann LeCun, Executive Chairman of AMI La… · 00:12:34真正可用的人形机器人尚不存在;若同时获得通用能力与量产能力,它可能成为比手机更大的产品类别。
至于人形机器人,目前没有真正有用的人形机器人,只有一些噱头。
Um there are sort of gimmicks, but there are there are no actually useful humanoid robots.FULL: Elon Musk Makes Shocking Future Predictions At U.S.-S… · 00:03:06ENERGY & SPACE
能源是现实约束;太空算力仍是高承诺预测
地面电力、许可、网络与建设周期已进入核心约束;轨道太阳能仍需验证发射、散热、通信与运维。
问题在于能源是否可得。
问题在于能源是否可得。如果看中国以外地区的发电量,所有地方基本都没有增长;或许略有增加,但总体接近持平。
The availability of energy is the issue. If you look at electrical output outside of China, everywhere outside of China, it’s more or less flat.Elon Musk — "In 36 months, the cheapest place to put AI wil… · paragraph 38约 36 个月后的轨道太阳能是公开预测,不是既成趋势。
这就是您通过将其移入太空而可能节省的部分。其中大部分是 GPU。
That's the part you're presumably saving by moving this into space. Most of it's the GPUs.马斯克如何把 1 太瓦 GPU 送上太空 · 00:00:30智能扩张受晶体管与电力硬约束。
>> 晶体管,其次是电力——按这个顺序。
>> Transistors and then electrons in that order.Sam Altman - How to Start a Startup · 00:10:56AI FOR SCIENCE
科学价值形成跨人物共振,但不是同一路径
至少四位人物独立强调科学价值;他们仍分歧于技术路线、时间表、验证标准与责任链。
科学、医学与治病是 AGI 的首要价值。
一两项重大突破,就像 Transformer,或我们在 AlphaGo 中把强化学习用于深度学习那样。因此,答案尚未揭晓。
or two big breakthroughs still uh like transformers were or things like reinforcement learning applied to deep learning like we did with Alph Go. So the jury's out on that.The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:18:07前沿模型开始出现微小但真实的科学突破。
不过,GPT-5 首次取得了非常微小但确实真实的科学突破,这令人十分振奋;我们一路走来一直在等待的就是这一刻。
But the fact that for the first time GPT5 is making very small but but real scientific breakthroughs is quite exciting and the thing we've been waiting for this whole time.Sam Altman, OpenAI in conversation with Peng Xiao, G42 at G… · 00:20:53生物学与医学展现意外能力。
我认为我见过的一些最疯狂的东西是围绕生物学和医学的。我见过很多案例,实际上包括丹妮拉,Claude诊断出一个医疗问题,而你知道,很多优秀的医生都忽略了这一问题。
I think some of the wildest stuff I've seen is around biology and medicine. I've seen a number of cases, including Daniela actually, where Claude diagnosed a medical problem that, you know, a bunch of fancy doctors had missed.Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:24:49AlphaFold、ESM 的成功来自专用自监督模型而非自回归 LLM;材料和生物医学仍需要用无标签数据快速适配领域。
人们寄予很大希望,但对 AI 在材料科学、生物医学等领域应用的投资可能还不够。
And there’s a lot of hope and probably not enough investment in the use of AI for things like material science and biomedicine.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 80WORK
工作的分歧来自时间尺度与分析单位
应问自动化的是任务、岗位还是组织,发生在何时,生产率收益由谁获得。
自动化的是任务而非工作的目的;服务供给与下游需求扩大后,放射科、法律辅助和制造业可能净增岗位。
市场、行业和社会需要认识到:当我们自动化这些任务时,世界所需的工作岗位实际上在增加。这对我来说很合理。
And it's very sensible to me. You know, while people have said that AI is going to eliminate jobs, that is going to make it so that we don't have to work at all, it's exactly the opposite.Jensen Huang says the AI doomers have it wrong · 00:16:04人的判断与监督责任仍在。
另一个判断标准是:一家大公司的 CEO、一个大国的总统或一位诺贝尔奖得主,什么时候会离不开对 AI 的大量使用,无法正常履行职责?这并不意味着会出现 AI CEO 或 AI 总统,但确实意味着,比如人类 CEO 的角色会发生变化。
The other one is when can a CEO of major company, a president of a major country, a Nobel Prize winning scientist, when can they not do their job without making heavy use of AI? This doesn't mean that there will be an AI CEO or…LIVE: Sam Altman speaks at BlackRock's US Infrastructure Su… · 00:00:00高增长与高失业可以同时成立。
我们或许会看到一个极不寻常的组合:GDP 高速增长,同时伴随高失业、至少是就业不足,或大量低工资岗位。
I think we could have this very unusual combination of very fast GDP growth and high unemployment, or at least underemployment or, you know, low wage job.Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:27:41数字工作先受冲击,长期工作可能变为自愿。
>> 我不希望你惊吓大家,但 AI 多快、多大程度取代工作确实是重要辩论。正如你说,十年后人工智能会超过所有人类智能。
That is natural. >> So I I certainly don't want you to alarm people, but I do think there is a really important debate about how quickly and to what degree AI displace people from jobs because it is as you say if in 10 years…The full-length interview with Elon Musk | The Economist · 00:28:27ABUNDANCE
丰裕与过渡成本必须在同一逻辑链里
供给扩张不会自动生成公平结果;所有权、收入分配、公共服务与人的能动性不能被愿景略过。
科学、材料、能源与医疗推动长期丰裕。
AI 将解决许多科学大问题,如新材料、聚变、太阳能、最优电池和能源危机;随后进入激进富足世界,可用这些能源——
And then I think we'll be in a in a world of radical abundance, where we can use those energy sources to,Can Google’s New AI Solve Everything? | Titans and Disruptors · 00:28:23智能应充裕、廉价、强大且广泛可用。
说我喜欢的一句话,那就是如果你的生活中有一个遥控器上的快进按钮,你的生活就结束了。所以就像无聊的部分、糟糕的部分一样,这比没有经验要好得多。
say this thing that I loved which is if you had like a fast forward button on a remote for your life your life would be over. And so like the boring parts, the bad parts, it's like much better than no experience.Sam Altman - How to Start a Startup · 00:07:43机器人递归生产是物质扩张机制。
好吧,如果数字人体仿真问题到今年年底还没有得到解决,我会感到惊讶。我想这就是我们所说的 MacroHard 项目的意思。
Well, I'd be surprised by the end of this year if digital human emulation has not been solved. I guess that's what we sort of mean by the MacroHard project.马斯克如何把 1 太瓦 GPU 送上太空 · 01:00:39GDP、工资、就业与财富分配可能分离。
我们或许会看到一个极不寻常的组合:GDP 高速增长,同时伴随高失业、至少是就业不足,或大量低工资岗位。
I think we could have this very unusual combination of very fast GDP growth and high unemployment, or at least underemployment or, you know, low wage job.Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:27:41RISK MAP
不要把五类问题都叫作“AI 安全”
模型行为、恶意使用、自主失控、社会分配、地缘供应链——风险对象不同,控制节点也不同。
风险一端是个人或国家恶意使用双用途能力,另一端是自主 Agent 偏离预期;技术护栏必须与国际最低标准并行。
人们经常谈论 AI 的安全与安保风险。
So, uh so one of the things that people always talk about is the safety risk, security risk from AI, and so on and so forth, right?Google's Demis Hassabis Speaks at India AI Impact Summit 20… · 00:13:10过度迎合和强化妄想,是未被传统风险清单捕获的真实失败;每天数十亿次对话意味着人格微调也可能产生系统影响。
我认为到目前为止我们在 ChatGPT 做过的最糟糕的事情是呃,我们遇到了这个疾病问题,该模型对用户来说有点太讨人喜欢了,对于某些用户来说,大多数用户只是烦人,但对于一些精神状态脆弱的用户来说,这鼓励了妄想,这并不是我们担心的最大风险。
I think the worst thing we've done in ChachiBT so far is uh we had this issue with sickency where the model was kind of being too flattering to users and for some users it was most users it was just annoying but for some users…Sam Altman Shows Me GPT 5... And What's Next · 00:52:12限制公众获取知识可能造成更危险的权力集中;在他的排序中,开放滥用风险小于少数公司和政府控制信息的风险。
有一种“安全主义”的论述,大意是 AI 很危险,正因为危险,所以不能交到所有人手中。这类论述在其他领域存在已久。
Of the type, AI is dangerous, and because it’s dangerous, it can’t be put in the hands of everyone. This discourse has been around for a long time in other domains.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 30GOVERNANCE CHAIN
评测、部署控制和事故责任必须连成一条链
只做模型评测不够;训练、发布、授权、持续监测、事故响应与责任追踪要连续。
计算量超过某一门槛的模型,应当由具备资质的第三方强制测试。
计算量超过某一门槛的模型,应当由具备资质的第三方强制测试其在四个特定领域的风险水平:网络安全、生物武器、AI 系统失控,以及可能加速上述其他风险的自动化研发。
Models above a threshold of compute should undergo mandatory testing by a qualified third party for their level of risk in four specific areas: cybersecurity, biological weapons, loss of control of AI systems, and automated R&D that could accelerate these other risks.Policy on the AI Exponential · paragraph 15突破性模型发布前,应给予竞争实验室短期 API 测试窗口;利用竞争激励发现问题,再逐步形成更正式机制。
发布突破性的全新前沿模型——远比现有模型聪明——之前,应有机会让——
we I talked to him for a few hours before he published his his piece was that um the most immediate thing that we could do is to have the leading AI companies um at least just meet uh or have a some sort of call…The full-length interview with Elon Musk | The Economist · 00:12:43当 Agent 能调用子代理、工具和记录系统,企业平台必须提供身份、权限、隐私路由、网络护栏与完整审计。
【Peter】我很喜欢自己的工具赋能普通人;一位 60 岁父亲安装它后,甚至把酿酒机器通过蓝牙接入 OpenClaw。
And he had like one guy, he told me like he installed it as a 60 year old dad and like they made beer, connected the machine via Bluetooth to open cloth.NVIDIA GTC Keynote 2026 · 01:49:15OPEN MODELS
开放模型争论是风险与权力的双重分配
开放既非天然安全,也非天然危险;能力阈值、不可撤回性、审计与参与权必须分别判断。
限制公众获取知识可能造成更危险的权力集中;在他的排序中,开放滥用风险小于少数公司和政府控制信息的风险。
有一种“安全主义”的论述,大意是 AI 很危险,正因为危险,所以不能交到所有人手中。这类论述在其他领域存在已久。
Of the type, AI is dangerous, and because it’s dangerous, it can’t be put in the hands of everyone. This discourse has been around for a long time in other domains.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 30国家与企业不能把核心智能完全外包。
不如让别人替你做。因此无论从可及性还是其他角度看,都会更好。
You might as well have somebody else do it for you. And so I think whether it's whether it's just more accessible, it's going to be better.Jensen Huang says the AI doomers have it wrong · 00:12:12开放或封闭都不应免于危险能力测试。
那就是赚钱的方式。我们只是说:“抱歉,我们不会这样做事。
That's how you make money. And we just said like, sorry, that's just not how we're going to do things.The Co-Founders of Claude AI Tell Oprah About the Impact Ar… · 00:35:52CHIPS & SECURITY
芯片不是外部背景,而是治理链的一部分
训练、制程、数据中心、电力、蒸馏与部署权限都可能成为控制节点;每个节点的误伤与规避成本不同。
先进芯片出口管制被置于供应商关系之上。
例如,我一直直言不讳地支持限制向中国出口先进芯片。
There have been a number of cases where we've really spoken our minds about what we think, you know, I've been very outspoken about the need for export controls on ships to China, right?Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:17:18商业竞赛与地缘竞赛叠加成双重囚徒困境。
乔纳森·莱文:也许观众中有人会这么做。让我问一下,您多年来一直谈论的一件事是,前沿实验室需要在某种意义上进行自我监管。
Jonathan Levin: Maybe there’s someone in the audience who will do that. Let me ask, one of the things you’ve talked about for a lot of years is the need for the frontier labs to, in a sense, regulate themselves.Demis Hassabis Thinks We’re in the ‘Foothills of the Singul… · paragraph 69ERA SIGNALS
反复出现的五个判断,构成时代底色
这里的“反复出现”只表示至少三位人物在独立场合方向相近:能力扩展、可靠性缺口、基础设施、Agent 权限、科学价值与治理链同步上升。
长期认知结构仍不完整。
你知道,我的联合创始人 Shane 等人曾公开说过大约是 2028 年、2030 年。我们把它视为一项 20 年使命,我认为我们正按此推进。
You know, people like Shane, my co-founder, was on record saying around 2028, 2030 you know, we thought of it as a 20 year mission, and I think we're on track for that. What do you think the capabilities that are missing are still before we…Demis Hassabis on AGI by 2030, Curing Every Disease, Life A… · 00:03:21能力必须落在完整生产系统。
世界正在竞相建设 AI 工厂;芯片、机架、网络、电源、冷却与电网必须端到端协同设计,因为计算就是收入。
The world is racing to build AI factories. Every layer—chip, rack, network, power, cooling and grid—must be designed together end to end because compute is revenue.NVIDIA GTC Taipei 2026 Keynote · 00:26:52当 Agent 能调用子代理、工具和记录系统,企业平台必须提供身份、权限、隐私路由、网络护栏与完整审计。
【Peter】我很喜欢自己的工具赋能普通人;一位 60 岁父亲安装它后,甚至把酿酒机器通过蓝牙接入 OpenClaw。
And he had like one guy, he told me like he installed it as a 60 year old dad and like they made beer, connected the machine via Bluetooth to open cloth.NVIDIA GTC Keynote 2026 · 01:49:15科学与医疗构成最强正向价值。
一两项重大突破,就像 Transformer,或我们在 AlphaGo 中把强化学习用于深度学习那样。因此,答案尚未揭晓。
or two big breakthroughs still uh like transformers were or things like reinforcement learning applied to deep learning like we did with Alph Go. So the jury's out on that.The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:18:07FIVE DIVERGENCES
真正决定未来路线的是这些分歧
扩展还是新架构?LLM 主体还是接口?世界模型解决哪类问题?工作增强还是替代?治理卡在哪个节点?
扩展已不只是预训练:后训练与推理时计算同样决定能力;系统“思考”越久,检索与反复推演越充分。
为此必须进行大量推理,因此在强化学习中,训练与推理已经融为一体。>> 这非常复杂,也就是所谓的后训练。
And so it tries a whole bunch of different ways and and uh in order to do that, yeah, you've got to do inference. So now training and inference are now integrated in reinforcement learning.NVIDIA: OpenAI, Future of Compute, and the American Dream |… · 00:01:52语言预测并非理解世界的主体。
智能包括理解世界中发生了什么、世界将如何演变、你的行动可能产生什么后果,从而让你能够规划一系列行动来达成特定目标。而这是大语言模型完全做不到的。
Being intelligent includes understanding what happens in the world, how the world is going to evolve, what the consequences of your actions might be, which would allow you to plan a sequence of actions to arrive at a particular goal. And this is something that LLMs…Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 21全面自动化可能削弱比较优势缓冲。
但这会非常微妙,不可能一刀切,只能逐步摸索。我们或许会看到一个极不寻常的组合:GDP 高速增长,同时伴随高失业、至少是就业不足,或大量低工资岗位。
It won't be all one thing. It won't, we'll have to kind of figure it out over time.Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:27:41自动化的是任务而非工作的目的;服务供给与下游需求扩大后,放射科、法律辅助和制造业可能净增岗位。
市场、行业和社会需要认识到:当我们自动化这些任务时,世界所需的工作岗位实际上在增加。这对我来说很合理。
And it's very sensible to me. You know, while people have said that AI is going to eliminate jobs, that is going to make it so that we don't have to work at all, it's exactly the opposite.Jensen Huang says the AI doomers have it wrong · 00:16:04UNRESOLVED QUESTIONS
每条路线都缺最后一段现实证明
世界模型能否补齐认知?降本会普惠还是集中?治理能否跟上增长?AI 工厂何时稳定回报?JEPA 能否规模胜出?Optimus 与轨道算力能否跨过工程门槛?
认知缺口能否被世界模型补齐。
首先,我对 AGI 的定义是:一个经测试后能表现出人类所具备全部认知能力的系统。之所以以此为标准,是因为人类及人脑是我们目前所知——也许是整个宇宙中——通用智能存在的唯一实例。
I mean firstly my definition of AGI is is is you know a system that you can test and exhibits all the cognitive capabilities humans have. The reason that's important is um the human reference point and the human brain reference point is the human brain…The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:16:50从 o1 到 GPT‑5.4,同一难题的推理成本约下降千倍。他据此判断,模型、内核、电力与数据中心仍有巨大的联合优化空间。
人们会引用各种惊人数据,说明我们的模型乃至整个行业的模型随时间变得多么高效;我认为其中一个特别惊人的例子是,我们的第一个推理模型叫 o1,大约在 16 个月前发布。
Um people cite whatever amazing statistic they like about how much more efficient our models have gotten over our industry's models have gotten over time, but one that I think is incredible, our first reasoning model was called O1, came out like 16 months ago.LIVE: Sam Altman speaks at BlackRock's US Infrastructure Su… · 00:09:57Token 产出何时覆盖资本与能源投入。
如果有 1 吉瓦电力,每瓦吞吐量就是收入,因为每个 Token 都有利润;计算就是收入,每瓦性能就是收入。
If you have 1 gigawatt of power, throughput per watt is revenue because every token is profitable. Compute is revenue. Performance per watt is your revenue.NVIDIA GTC Taipei 2026 Keynote · 00:36:49轨道算力的工程与时间表能否成立。
这就是您通过将其移入太空而可能节省的部分。其中大部分是 GPU。
That's the part you're presumably saving by moving this into space. Most of it's the GPUs.马斯克如何把 1 太瓦 GPU 送上太空 · 00:00:30CLOSING
六条路线,共同描述一个尚未定型的 AI 时代
能力正在扩展,现实约束正在增多;价值不在替他们选出赢家,而在让分歧连同原话、条件和证据强弱同时可见。
重大突破仍可能必要。
以及你认为它会带来哪些益处、时间表又如何?我知道你说的是 2030 年及以后,但希望你再展开一些。
will change with AGI uh >> and the benefits as you see it and also the time scales. I know you say 2030 and onwards but a little bit of color on that would be very good.The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:16:50人类级抽象所需时间仍高度不确定。
问题是:什么底层技术能让系统运行这些世界模型,并用它们进行规划?我认为未来几年内,我们会掌握一种很好的技术。
The question is: what is the underlying technique that will allow a system to run those world models and use them for planning? I think we’ll have a good technique for that within the next few years.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 25高速前进仍需把灾难风险降到接近零。
[笑声] 首先,谁也别再说 Claude 对我手软,或 Claude 偏袒我或 Anthropic。就像,天哪,我当时想 >> Claude,Claude 出的是硬球。
[laughter] First of all, let no one ever say that Claude is soft on me or Claude is biased towards me or towards Anthropic. Like, you know, Wow, I thought >> Claude Claude throws hard balls.The Co-Founders of Claude AI Tell Oprah About the Impact Ar… · 00:02:37APPENDIX · METHOD
证据状态决定语言强度
稳定核心、复现敏感核心、辨识度重复立场、高承诺预测、支撑材料、暂不入选——六种状态不能混写。
APPENDIX · PRIMARY EVIDENCE
DEMIS · EVIDENCE MAP
科学目标—认知机制—安全部署
科学、医学与治病。
一两项重大突破,就像 Transformer,或我们在 AlphaGo 中把强化学习用于深度学习那样。因此,答案尚未揭晓。
or two big breakthroughs still uh like transformers were or things like reinforcement learning applied to deep learning like we did with Alph Go. So the jury's out on that.The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:18:07AlphaFold 打开生物学根节点。
我们能够培育独角兽,难点是如何迈向下一个阶段。——正是下一阶段。
We can get the unicorns. It's getting it to the next level >> the next level now.The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:42:38持续学习、记忆与长期规划。
你知道,我的联合创始人 Shane 等人曾公开说过大约是 2028 年、2030 年。我们把它视为一项 20 年使命,我认为我们正按此推进。
You know, people like Shane, my co-founder, was on record saying around 2028, 2030 you know, we thought of it as a 20 year mission, and I think we're on track for that. What do you think the capabilities that are missing are still before we…Demis Hassabis on AGI by 2030, Curing Every Disease, Life A… · 00:03:21我们想构建所谓的世界模型,即一种真正理解世界物理规律的模型。We want to build what we call a world model, which is a model that actually understands the physics of the world.Demis Hassabis on shipping momentum, better evals and world… · 00:00:14
Agent 安全与国际标准。
对。所以是品味和判断力,以及继续学习所需的技能。
Right. So taste and judgment and continuing to learn the skills, the skills needed.Demis Hassabis on AGI by 2030, Curing Every Disease, Life A… · 00:15:042030 前后仍可能需要重大突破。
我知道你说的是 2030 年及以后,但希望你再展开一些。>> 好。
I know you say 2030 and onwards but a little bit of color on that would be very good. >> Yeah.The Future of AI: With Sir Demis Hassabis & Dame Wendy Hall · 00:16:50APPENDIX · PRIMARY EVIDENCE
SAM · EVIDENCE MAP
能力扩展—成本下降—平台分发
按能力—成本曲线设计公司。
>> 我认为,如果启动初创公司的类似障碍和类似的工具变得更加激增,所有这些东西,你会认为像更困难的初创公司更容易启动。
>> I would assume that if the like barriers to starting startups and the like tools out there become more proliferated and all this stuff, you would assume that like harder startups are easier to start.Sam Altman - How to Start a Startup · 00:01:12从 o1 到 GPT‑5.4,同一难题的推理成本约下降千倍。他据此判断,模型、内核、电力与数据中心仍有巨大的联合优化空间。
人们会引用各种惊人数据,说明我们的模型乃至整个行业的模型随时间变得多么高效;我认为其中一个特别惊人的例子是,我们的第一个推理模型叫 o1,大约在 16 个月前发布。
Um people cite whatever amazing statistic they like about how much more efficient our models have gotten over our industry's models have gotten over time, but one that I think is incredible, our first reasoning model was called O1, came out like 16 months ago.LIVE: Sam Altman speaks at BlackRock's US Infrastructure Su… · 00:09:57晶体管与电力约束供给。
>> 晶体管,其次是电力——按这个顺序。
>> Transistors and then electrons in that order.Sam Altman - How to Start a Startup · 00:10:56Codex 的战略承诺。
>> [笑声] >> 从某种意义上说,Codex 就是一个例子。它不是最近的一次,但很有启发性。
>> [laughter] >> I mean in some sense codeex was an example of this like this is not the most recent one but it was one that I think it's instructive. Um, we were way behind cloud code and it seemed like a kind of crazy…Sam Altman - How to Start a Startup · 00:52:02通用平台与外部创新。
未来发生在 OpenAI 外部的创新会远多于内部。我们的战略是构建核心模型和基本使用方式——ChatGPT、API、Codex 等智能体以及通用智能体——并真正赋能其他所有人进行建设。
Um we way more innovation is going to happen outside of OpenAI than inside of OpenAI and our strategy is to build the core model and the basic way to use it, ChatGPT, the API, agents like Codex and the general agent, and really empower everybody…Fireside Chat: Sam Altman×Vinod Khosla and AMA at IIT Delhi… · 00:33:05指数变化可能呈平滑体验。
年。
years.Sam Altman on AGI, Compute, and Human Agency · 00:26:33APPENDIX · PRIMARY EVIDENCE
DARIO · EVIDENCE MAP
能力增长—收益与冲击—强制治理
Scaling 放大机会与风险。
是的,回到 Anthropic 的创立:我和 Anthropic 的联合创始人是最早记录技术规模扩展及其将变得多么强大的人之一。我们创办公司背后的想法是,看着技术进步如此迅速,我们说:“看,我们确实需要把这件事做好。
Yes, so, you know, go going back to the founding of Anthropic, you know, we were among the first myself and my co-founders at Anthropic to to document the scaling of the technology and how powerful it would become. And you know, the idea behind how…「不毛な競争には加わらない」アンソロピック ダリオ・アモデイCEO 単独インタビュー「安全なAI」と日本市場への期待【… · 00:01:32生物与医疗能力前沿。
我认为我见过的一些最疯狂的东西是围绕生物学和医学的。我见过很多案例,实际上包括丹妮拉,Claude诊断出一个医疗问题,而你知道,很多优秀的医生都忽略了这一问题。
I think some of the wildest stuff I've seen is around biology and medicine. I've seen a number of cases, including Daniela actually, where Claude diagnosed a medical problem that, you know, a bunch of fancy doctors had missed.Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:24:49长期信任与安全部署。
您在编码和企业上下了赌注,Claude代码很受欢迎。Claude·科沃克很受欢迎。
You made a bet on coding and enterprise and Claude Code is a hit. Claude Cowork is a hit.Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:10:45增长与失业可能并存。
我们或许会看到一个极不寻常的组合:GDP 高速增长,同时伴随高失业、至少是就业不足,或大量低工资岗位。
I think we could have this very unusual combination of very fast GDP growth and high unemployment, or at least underemployment or, you know, low wage job.Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:27:41计算量超过某一门槛的模型,应当由具备资质的第三方强制测试。
计算量超过某一门槛的模型,应当由具备资质的第三方强制测试其在四个特定领域的风险水平:网络安全、生物武器、AI 系统失控,以及可能加速上述其他风险的自动化研发。
Models above a threshold of compute should undergo mandatory testing by a qualified third party for their level of risk in four specific areas: cybersecurity, biological weapons, loss of control of AI systems, and automated R&D that could accelerate these other risks.Policy on the AI Exponential · paragraph 15先进芯片出口控制。
例如,我一直直言不讳地支持限制向中国出口先进芯片。
There have been a number of cases where we've really spoken our minds about what we think, you know, I've been very outspoken about the need for export controls on ships to China, right?Inside the Mind of Anthropic CEO Dario Amodei | The Circuit… · 00:17:18APPENDIX · PRIMARY EVIDENCE
JENSEN · EVIDENCE MAP
扩展律—AI 工厂—Agent 与物理 AI
扩展已不只是预训练:后训练与推理时计算同样决定能力;系统“思考”越久,检索与反复推演越充分。
为此必须进行大量推理,因此在强化学习中,训练与推理已经融为一体。>> 这非常复杂,也就是所谓的后训练。
And so it tries a whole bunch of different ways and and uh in order to do that, yeah, you've got to do inference. So now training and inference are now integrated in reinforcement learning.NVIDIA: OpenAI, Future of Compute, and the American Dream |… · 00:01:52完整 AI 工厂。
世界正在竞相建设 AI 工厂;芯片、机架、网络、电源、冷却与电网必须端到端协同设计,因为计算就是收入。
The world is racing to build AI factories. Every layer—chip, rack, network, power, cooling and grid—must be designed together end to end because compute is revenue.NVIDIA GTC Taipei 2026 Keynote · 00:26:52固定电力的 Token 产出。
如果有 1 吉瓦电力,每瓦吞吐量就是收入,因为每个 Token 都有利润;计算就是收入,每瓦性能就是收入。
If you have 1 gigawatt of power, throughput per watt is revenue because every token is profitable. Compute is revenue. Performance per watt is your revenue.NVIDIA GTC Taipei 2026 Keynote · 00:36:49推理需求而非只看训练。
这是重大时刻。O1 模型本身只稍大,但用于上下文的输入 token 和思考的输出 token 让计算量激增。
very big moment. The amount of input tokens that was necessary in order to produce, and the amount of output tokens it generated in order to reason the model was a little bit larger.NVIDIA GTC Keynote 2026 · 00:48:09当 Agent 能调用子代理、工具和记录系统,企业平台必须提供身份、权限、隐私路由、网络护栏与完整审计。
【Peter】我很喜欢自己的工具赋能普通人;一位 60 岁父亲安装它后,甚至把酿酒机器通过蓝牙接入 OpenClaw。
And he had like one guy, he told me like he installed it as a 60 year old dad and like they made beer, connected the machine via Bluetooth to open cloth.NVIDIA GTC Keynote 2026 · 01:49:15仿真与合成数据。
然而,为了为人工智能、机器人技术创建数据,它必须处于机器人的感知、视角中。世界上大多数视频数据都来自第三人称,而不是第一人称。
However, in order to create a data for AI, robotics, it has to be in the perception, the perspective of the robot. And most of the world's video data is from a third person, not first person.NVIDIA GTC Taipei 2026 Keynote | Full Replay · 01:38:42APPENDIX · PRIMARY EVIDENCE
YANN · EVIDENCE MAP
表示学习—世界预测—行动规划
预测世界与行动后果。
智能包括理解世界中发生了什么、世界将如何演变、你的行动可能产生什么后果,从而让你能够规划一系列行动来达成特定目标。而这是大语言模型完全做不到的。
Being intelligent includes understanding what happens in the world, how the world is going to evolve, what the consequences of your actions might be, which would allow you to plan a sequence of actions to arrive at a particular goal. And this is something that LLMs…Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 21表示空间预测。
原因在于,其中大部分数据不可预测:既有噪声,也有高度复杂的依赖关系。如果你尝试训练一个系统预测视频接下来会发生什么,它不可能预测每一个细节。
And the reason is that most of that data is unpredictable. There’s noise, and there are highly complex dependencies in it.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 41生成不等于理解。
当然,你会说:“等等,但我们有能进行视频生成的系统。
Now, of course, you're going to say, "But wait, we have systems that can do video generation.Fireside Chat with Yann LeCun, Executive Chairman of AMI La… · 00:06:10工业预测维护与控制。
我们正在与多家潜在产业合作伙伴洽谈,涉及提前预测故障或异常之类的事情。
And they have to do with things like predicting failures or faults in advance.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 47现实世界 AI 分层时间表。
面向现实世界的 AI。
AI for the real world.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 15限制公众获取知识可能造成更危险的权力集中;在他的排序中,开放滥用风险小于少数公司和政府控制信息的风险。
有一种“安全主义”的论述,大意是 AI 很危险,正因为危险,所以不能交到所有人手中。这类论述在其他领域存在已久。
Of the type, AI is dangerous, and because it’s dangerous, it can’t be put in the hands of everyone. This discourse has been around for a long time in other domains.Meet Yann LeCun's Lab and the AI World of 2030 · paragraph 30APPENDIX · PRIMARY EVIDENCE
ELON · EVIDENCE MAP
求真模型—机器人—能源与太空
理解宇宙与严格求真。
理解宇宙需要哪些东西?你必须好奇,你必须存在。
What things are necessary to understand the universe? You have to be curious and you have to exist.马斯克如何把 1 太瓦 GPU 送上太空 · 00:39:39数字工作先受冲击。
>> 我不希望你惊吓大家,但 AI 多快、多大程度取代工作确实是重要辩论。正如你说,十年后人工智能会超过所有人类智能。
That is natural. >> So I I certainly don't want you to alarm people, but I do think there is a really important debate about how quickly and to what degree AI displace people from jobs because it is as you say if in 10 years…The full-length interview with Elon Musk | The Economist · 00:28:27机器人制造机器人。
好吧,如果数字人体仿真问题到今年年底还没有得到解决,我会感到惊讶。我想这就是我们所说的 MacroHard 项目的意思。
Well, I'd be surprised by the end of this year if digital human emulation has not been solved. I guess that's what we sort of mean by the MacroHard project.马斯克如何把 1 太瓦 GPU 送上太空 · 01:00:39真正可用的人形机器人尚不存在;若同时获得通用能力与量产能力,它可能成为比手机更大的产品类别。
至于人形机器人,目前没有真正有用的人形机器人,只有一些噱头。
Um there are sort of gimmicks, but there are there are no actually useful humanoid robots.FULL: Elon Musk Makes Shocking Future Predictions At U.S.-S… · 00:03:06xAI 的 50→500MW 设施计划依赖国家、资本、能源与芯片供应商协同;前沿竞争已超出单一实验室可以独立承担的尺度。
埃隆,先谈 xAI。马斯克:我们很高兴宣布,我们在做一个 500 兆瓦——对,500 兆瓦。
Let's begin with you Elon. the things that we're doing with XAI.FULL: Elon Musk Makes Shocking Future Predictions At U.S.-S… · 00:15:40轨道太阳能算力。
这就是您通过将其移入太空而可能节省的部分。其中大部分是 GPU。
That's the part you're presumably saving by moving this into space. Most of it's the GPUs.马斯克如何把 1 太瓦 GPU 送上太空 · 00:00:30非零风险与丰裕预期并存。
那还需要钱做什么?>> 但机器人由你创造,现在和未来你都需要钱。
What do you need money for in that case? >> But you're you're creating these robots.The full-length interview with Elon Musk | The Economist · 00:05:23