忆往昔，符号AI曾以规则逻辑统领江湖；今朝卷土重来，它携手神经网络，直指AGI！ 但AI领域的权威们已经开始泼下一盆冷水： 真正的突破，恐怕要靠老牌选手「符号派AI」与神经网络联手登场。 这几年，大模型多次让人惊艳：聊天像真人、写作像专家、画画像大师，仿佛「万能AI」真的要来了。 只靠「神经网络」，远远不够通往人类级智能。 美国人工智能促进协会（AAAI）向会员发出提问： 绝大多数研究者给出的答案是——不行。 符号AI：起死回生 在历史上，符号派AI曾是主角——它相信，世界可以被规则、逻辑和清晰的概念关系穷尽刻画： 像数学那样精确，像流程图那样可追溯，像生物分类法那样层次分明。 后来，神经网络崛起，用「从数据中学习」的范式席卷整个领域。 大模型与ChatGPT成为这个时代的技术图腾，而符号系统被边缘化，几乎只剩下教科书上的一段历史。 然而，自2021年前后开始，「神经–符号融合」急速升温，被视为打破单一神经网络话语权的一次反扑： 未来，计算机能否达到、甚至超越人类智力？ 如果可以，单靠当下火爆的神经网络行不行？ 它试图把统计学习与显式推理拼接在一起，不仅为了追逐通用智能这一远目标，更为了在军事、医疗等高风险场 ...

自 8 月 GPT-5 发布以来， Codex 展现出惊人的爆发力， 用户增长 20 倍 ，每周处理数万亿 tokens，成为了 Open AI 最受欢 迎的编程智能体。 "Codex 能快速实现 20 倍 的增长，不只是因为模型变强了，还因为我们理解了，真正的智能体不是一个模型，而是模型、 API 和框架共同努力的结果。" 在最新播客中，OpenAI 的编程智能体 Codex 产品负责人 Alexander Embiricos 揭露背后的秘 密。 比如，Codex 在长时任务能力上的突破。为了让它能够连续工作十几个小时甚至数天，团队设计了名为"压缩"的机制——模型 负责提炼关键信息，API 承接任务链路，框架负责稳定运行。三层像齿轮般咬合，使 Codex 能够完成传统大模型难以支撑的 长时编程任务。 正是这样的底层逻辑，让 Codex 在业务实战中有惊人表现。 作者 | 高允毅 Andrej Karpathy 曾公开分享，他被一个 bug 困住数小时，最终交给 Codex 处理，一小时内就完成了修复。 Sora 团队更是依靠 Codex，在短短 28 天时间，从 0 到 1 完成 Android 应用的上线 ...

In a heartfelt blog celebrating ChatGPT-maker OpenAI's 10th anniversary, CEO Sam Altman reflected on a “decade of breakthroughs, learnings, and the path toward AGI that benefits all of humanity”.The post by Sam Altman reminisced the artificial intelligence platform's turbulent start and journey towards success over the past 10 years. “OpenAI has achieved more than I dared to dream possible; we set out to do something crazy, unlikely, and unprecedented. From a deeply uncertain start and against all reasonabl ...

Software agents now plan travel, shop online, and negotiate subscriptions; the next step extends that autonomy from clicks to settlement, since a wallet can be created in code and funded without manual steps. That shift recasts payments as an API call, and it places public chains and stablecoins in the centre of a new transaction layer that never sleeps. The idea is not science fiction for distant horizons; it follows directly from how agents already fetch data, route tasks, and make bounded choices, whic ...

Why London. We have a huge pipeline, actually, by the way. Thanks for having me.I'm very excited to be here. We have a huge pipeline of of researchers, engineers from Europe and also DeepMind that want to join Luma. And also, London is the gateway for the business in Europe as well as Middle East.So it seems to be the right place, too, to have our second office outside of Palo Alto in the area. So, yeah, today is our launch of the London office. I have a feeling Demis Hassabis over at DeepMind and the labs ...

It's going to be wonderful for entrepreneurship. So you Harry I think are going to be very happy today. For a lot of ideas founders are intelligence constrained. I think of being capital constrained as a form of being intelligence constrained.For example, if you pick a therapist who wants to start a mental health startup today that founder would have to raise at least a few hundredk if not a few million to recruit some software engineers, maybe a marketing person or a growth person, maybe a product manager. ...

一场惊天逆袭。 编辑/艾伦 来源/新智元 (ID：AI_era) 估值飙升至3 5 0 0亿美元，Cl a u d e强势登顶企业级市场，An t h r o p i c正式终结了Op e nAI 的 独 角 戏 时 代 。 从 I n st a g r am 创 始 人 到 Op e nAI 的 「 决 裂 者 」 ， 这 11 位 顶 尖 人 物 组 成 了 硅谷最豪华的「复仇者联盟」。他们用一场惊天逆袭证明，在通往AGI的狂飙突进中， 对安全的极致坚守才是最深的护城河。 如果你在2 0 2 1年走进An t h r o p i c 的办公室，可能只会看到一群从Op e nAI「叛逃」出来 的理想主义者； 但如果你今天再看An t h r o p i c的高管名单，你会发现这已经不再是一个简单的实验室， 而是一支足以撼动硅谷版图的「全明星复仇者联盟」。 随着An t h r o p i c的估值在本月飙升至惊人的3 5 0 0亿美元，以Cl a u d e Op u s 4 . 5为代表的 Cl a u d e系列，更是拿下了企业级市场3 2 %的份额。 在这个庞大的数字帝国背后，是11个性格迥异、背景传 ...

A Bitcoin bounceback? Tether troubles? Fed fluctuations? These and more occupy this edition of Web3 Thoughts of the Week.Fed response key to Bitcoin bounceback“The scale of this pullback is the market’s response to uncertainty, not a collapse in underlying demand. Many investors have been reducing exposure because they lack clarity on the Federal Reserve’s next move, among other reasons. Once that clarity arrives, positioning will likely shift quickly.“More than $19 billion in long positions have already be ...

Group 1 - The core focus of the article is on Elon Musk's recent interview discussing advancements in Neuralink, xAI, and Grok 5, highlighting his vision for AGI and the future of technology [2][3][4] - Neuralink aims to evolve from basic communication to creating "bionic superhumans," with the potential to enable paralyzed individuals to run faster than normal humans at a cost of approximately $60,000 [5][9] - Musk's acquisition of Twitter, now X, was driven by a desire to create a platform for free speech, viewing it as a "global town square" rather than a traditional social media platform [10][16][17] Group 2 - Musk's departure from OpenAI was due to a fundamental disagreement over its direction, leading him to establish xAI, which focuses on leveraging unique data and hardware capabilities to compete in the AGI space [18][20][21] - Grok 5 is presented as a significant advancement in AI, with 6 trillion parameters and a 10% probability of achieving AGI, marking a shift in Musk's perspective on the feasibility of AGI [22][25][26] - The development of GrokPedia aims to create an open-source repository of human knowledge, with plans to back it up in space, emphasizing the importance of preserving knowledge for future generations [26][27] Group 3 - Tesla's manufacturing vision includes reducing vehicle production time to 5 seconds per car, likening the factory to a "giant GPU" and emphasizing the need for rapid innovation in manufacturing processes [33][34][36] - Musk's strategy involves developing in-house chip manufacturing capabilities to meet Tesla's growing demand, as existing suppliers cannot scale quickly enough to support the company's ambitious goals [37][40][41] - The full self-driving (FSD) technology is reported to be significantly safer than human driving, with Musk aiming for a tenfold increase in safety with future iterations of the AI system [42][45]

Core Insights - Microsoft has entered a significant partnership with Anthropic, positioning itself as a major player in the AI sector for the long term [1] - Anthropic will purchase $30 billion worth of Azure's computing capacity from Microsoft, which will enhance the capabilities of its Claude AI model [2] - The deal is substantial for Microsoft but is not its most critical AI agreement, which remains with OpenAI [2][4] Partnership Details - The agreement allows Anthropic to scale its Claude AI model on Azure, utilizing Nvidia's architecture powered by Grace CPUs and Blackwell and Vera Rubin GPUs [2] - Claude AI's availability on the three largest platforms will provide it with a competitive edge and increase its popularity, especially with the introduction of Claude for Business on Microsoft Foundry [3] Market Position - Microsoft has a market capitalization of $3.5 trillion and has been publicly traded since 1986, making this one of its largest deals [4] - The relationship with OpenAI, established in 2019, is more significant, with Microsoft holding a 27% stake valued at approximately $135 billion [5] Future Commitments - OpenAI is contracted to purchase $250 billion worth of Azure services through 2032, significantly exceeding the $30 billion deal with Anthropic [6] - The recent evolution of the agreement with OpenAI allows Microsoft to independently pursue partnerships in the AI space, enhancing its flexibility [8] Technological Advancements - Microsoft will provide computational resources for both Anthropic's Claude AI and OpenAI's ChatGPT, establishing Azure as a critical cloud ecosystem for AI [9] - The agreements with both companies position Microsoft to leverage advancements in artificial general intelligence (AGI), which could surpass human cognitive capabilities [10] Strategic Implications - These partnerships enable Microsoft to benefit from innovations driven by two leading AI models and offer the flexibility to collaborate more with Anthropic as its offerings evolve [11] - While the OpenAI partnership remains paramount, the Anthropic deal solidifies Microsoft's status as a key player in the AI industry [11]