引领具身智能全球创新。 本文为IPO早知道原创 作者｜ Stone Jin 微信公众号｜ipozaozhidao 据 IPO早知道消息， 智元机器人 日前 与国际顶尖具身智能公司 Physical Intelligence（Pi）携 手，双方将围绕动态环境下的长周期复杂任务，在具身智能领域展开深度技术合作。 目前， 智元机器人与 Pi的合作已经初具成效，可以实现一个通用模型根据不同的指令输入执行多个 任务，也可以适配多种末端执行器，包括灵巧手和夹爪，同时可以兼容鱼眼和针孔相机等多种传感 器。 Pi 作为 全球具身智能技术领导者，专注于将通用人工智能（ AGI）技术应用于现实物理世界，由包 括具身智能领域先驱Sergey Levine, Chelsea Finn教授在内的全球顶尖科学家、工程师、机器人 学者共同创立，研发了π0、Hi Robot等先进具身模型。 智元机器人致力以 AI+机器人的融合创新，打造世界级领先的通用具身机器人产品及应用生态。智元 机器人构建了领先的机器人"本体+AI"全栈技术，在具身智能领域拥有本体-数据-模型三位一体全栈 布局，量产下线超过1000台通用具身机器人。 此外， 罗剑岚博士 ...

星标 ★ IPO日报 精彩文章第一时间推送 近日，国内AI语音独角兽——云知声智能科技股份有限公司（以下简称"云知声"）向港交所主板提交上市申请，中金公司和海通国际为联 席保荐人。 云知声自2012年成立以来，一直致力于通过通用人工智能（AGI）技术创建互联直觉的世界。作为一家以技术创新为核心驱动力的公司，云知 声在语音识别、自然语言处理、机器学习等领域不断深耕，致力于将人工智能技术转化为实际应用，为各行各业提供智能化解决方案。 目前来看，云知声在技术突破和市场拓展方面取得了显著进展，营收也保持了稳步增长，但如何盈利可能是它的挑战——高额的研发投入是一 个方面，而与此同时，越来越多科技巨头和初创企业涌入人工智能领域，市场竞争激烈程度在不断加剧。 制图： 佘诗婕 三位"75后"博士联手创业 云知声的故事始于三位"75后"博士——梁家恩、黄伟和康恒联手创业。 具体来看，梁家恩，现年48岁，为云知声共同创办人、董事长、执行董事、副总经理兼首席技术官。曾2001年7月获得中国安徽省中国科学技 术大学自动控制专业学士学位，并于2006年7月获得中国北京市中国科学院自动化研究所模式识别与智能系统专业博士学位。 黄伟，现年 ...

图片来源： DeepMind "如今再想公开发表像Transformer那样的论文，几乎是不可能的。"一位现任研究员感慨道。 据两位知情人士透露，DeepMind对涉及生成式AI的"战略性"论文设置了长达六个月的"冷却期"，研 究人员还需说服多个团队，证明论文的发表价值。 一位接近公司的人士解释称，此举旨在减少研究 人员在"战略或商业上难以获批"的项目上浪费时间。他补充说，DeepMind每年仍发表数百篇论文， 并持续在顶级AI会议上保持影响力。 据七位现任及前任DeepMind研究人员透露，公司已引入更严格的审查机制和行政流程，显著提高了 论文发表的门槛。三位前员工表示， DeepMind尤其避免公开可能被竞争对手利用的技术创新，或可 能使谷歌Gemini大模型在对比中处于劣势的研究。 这一转变标志着DeepMind的战略重心从学术影响力向商业竞争力的倾斜。过去，该机构以发表突破 性论文和吸引全球顶尖AI人才著称——2017年谷歌研究人员发布的Transformer论文奠定了大语言模 型的基础，并直接推动了生成式AI的爆发。然而，随着OpenAI等竞争对手的崛起，DeepMind已成为 谷歌重振AI领导地位 ...

本报赴博鳌特派记者 陈子帅 在今年的博鳌亚洲论坛年会上，人工智能（AI）无疑是焦点话题之一。从技术创新到应用前景，再到 全球治理与合作，与会嘉宾们在思想交锋中探讨着AI发展的未来之路。作为论坛"AI界大咖"之一，中国 工程院院士、清华大学智能产业研究院（AIR）院长张亚勤在现场接受《环球时报》记者专访时表示， AI已经开启第四次工业革命，中国有机会成为这场智能革命的领军者。而未来10年，机器人的普及将 改变人类的生活生产方式。 实现这一场景的前提是机器人"大脑"的升级。张亚勤说，未来机器人的后台——"大脑"的80%都是相同 的，只是前台的表现形式——"四肢"不同。"与现在相比，这是一个巨大的改变。现在每个大模型、每 个特定任务的数据都不同，只有未来机器人拥有共同的'大脑'，才能实现智能涌现的本质。" 当机器人数量超越人类后，我们的生活和生产方式将会出现重大改变吗？"这意味着生产力的大幅提 升，人们有了更多选择的自由。"张亚勤表示，200多年前工业革命的影响至今犹在——它使人们在工作 岗位上、在工厂流水线上付出大量时间，机器人革命有望改变这一情况——那些"必须性工作"将会减 少，10年后人们可能每周只工作两三天 ...

Core Insights - Artificial Intelligence (AI) is a focal topic at the Boao Forum, with discussions on its technological innovations, application prospects, and global governance [1] - Zhang Yaqin, a prominent figure in AI, predicts that robots will outnumber humans in the next decade, significantly altering human lifestyles and production methods [2][3] Industry Trends - The next ten years will see a substantial increase in robot adoption, with costs comparable to smartphones, making them accessible to everyone [2] - Zhang Yaqin envisions that individuals may own an average of ten robots, which will serve various roles such as companions, secretaries, and drivers [2][3] Technological Development - Achieving advanced AI capabilities requires a significant upgrade in the "brains" of robots, with 80% of their backend being similar, differing only in their physical forms [2][3] - The path to achieving Artificial General Intelligence (AGI) is outlined in three stages: information intelligence, physical intelligence, and biological intelligence, with a timeline of 15 to 20 years [4] Economic Impact - The anticipated robot revolution could lead to a drastic reduction in mandatory work hours, potentially allowing people to work only two to three days a week [3] - The fourth industrial revolution, driven by AI, presents an opportunity for China to become a leader, contrasting its previous roles in earlier industrial revolutions [5] Global Cooperation and Risks - The forum highlighted the importance of balancing AI application and governance, emphasizing the need for international consensus and cooperation, especially for developing countries [6][7] - Zhang Yaqin warns against "AI isolation," where different countries develop separate AI systems, which could hinder global progress and exacerbate inequalities [7]

Core Viewpoint - The article discusses the emergence of Multimodal Chain of Thought (MCoT) as a significant advancement in AI, enabling it to process and reason across various modalities such as images, audio, and text, thereby enhancing its reasoning capabilities to be more human-like [1][4][17]. Summary by Sections MCoT Overview - MCoT represents a shift from traditional Chain of Thought (CoT) by integrating multiple sensory inputs, allowing AI to perform complex reasoning tasks that reflect real-world scenarios [2][3][4]. - The development of MCoT is a collaborative effort from researchers at several prestigious institutions, addressing the lack of comprehensive reviews in this field [5]. MCoT Methodology - MCoT's success relies on a systematic methodology comprising six technical pillars, enhancing the precision and fluency of academic expression [7]. 1. Reasoning Construction Perspective - Prompt-based: Utilizes carefully designed multimodal instruction templates to guide models in generating reasoning chains in few-shot scenarios [8]. - Plan-based: Constructs dynamic reasoning paths, allowing models to explore multiple hypotheses and select optimal solutions [8]. - Learning-based: Embeds reasoning tasks during training to enhance the model's intrinsic reasoning capabilities [8]. 2. Structured Reasoning Perspective - Asynchronous Modality Modeling: Decouples perception and reasoning modules to improve modular efficiency [10]. - Defined Procedure Staging: Employs predefined procedural rules to ensure orderly reasoning processes [10]. - Autonomous Procedure Staging: Dynamically generates sub-task sequences based on task requirements [10]. 3. Information Enhancement Perspective - Expert Tools Integration: Combines specialized tools to improve task accuracy and practicality [12]. - World Knowledge Retrieval: Utilizes retrieval-augmented generation techniques to enrich model background information [12]. - In-context Knowledge Retrieval: Analyzes entity relationships within task contexts to enhance logical consistency [12]. 4. Target Granularity Perspective - Introduces multimodal thinking processes to improve interpretability and intuitiveness in reasoning tasks [14]. - Coarse Understanding: Focuses on macro-level scene understanding [14]. - Semantic Grounding: Achieves mid-level analysis by detecting specific object locations [14]. - Fine-grained Understanding: Conducts micro-level analysis for precise segmentation [14]. 5. Multimodal Rationale - Emphasizes the importance of reasoning across multiple modalities to enhance AI's cognitive capabilities [15]. 6. Testing and Expansion Perspective - Slow-Thinking Mechanism: Encourages deep reasoning through long-chain examples and diverse reasoning paths [16]. - Reinforcement Learning Optimization: Guides reasoning processes with reward functions to improve performance in complex tasks [16]. Applications and Future Challenges - MCoT is already influencing various sectors, including robotics, autonomous driving, healthcare, creative generation, and education [17][25]. - Key challenges for MCoT's future development include: 1. Efficient use of computational resources, requiring algorithm improvements and hardware optimization [18][19]. 2. The chain effect of reasoning errors, necessitating real-time error detection and correction algorithms [20][21]. 3. Ethical concerns regarding content credibility, prompting the need for content verification frameworks [22][23]. 4. The diversity of task scenarios, which calls for cross-domain evaluation systems to enhance MCoT's applicability [24].

以下文章来源于凤凰网科技 ，作者凤凰网科技 凤凰网科技 . 凤凰科技频道官方账号，带你直击真相。 将投中网设为"星标⭐"，第一时间收获最新推送 行业当前对Anuttacon的争议，主要聚焦在其能有多大的创新，这决定了蔡浩宇能不能匹配上"下一 个梁文锋"的称号。 作者丨董雨晴 来源丨凤凰网科技 "这是紧急求救信号""来自盖亚星球的广播""如果你收到这个，请立即回复"。 2025年3月，一款科幻题材互动游戏在X平台上发布了部分片段，很快引发了讨论。 根据预告，玩家会在这款游戏中体验一种前所未有的AI驱动角色玩法，其核心机制是以实时对话推 动剧情发展，玩家的任务是帮在外星球上的女主角Stella找到回家的路。 AI游戏本不新鲜，新鲜的是，这款游戏背后的掌舵者，是米哈游创始人蔡浩宇。 凤 凰 网 科 技 了 解 到 ， 去 年 9 月 ， AI 行 业 收 到 了 一 条 英 雄 帖 。 蔡 浩 宇 二 次 创 业 创 办 的 新 公 司 Anuttacon，向行业广纳精英，重点招聘预训练与LLM人才，办公地点则是在硅谷。 一位接触过Anuttacon的投资行业人士在当时向投资界表示，Anuttacon早期是做AI+ ...

责编 | 王启隆 出品丨AI 科技大本营（ID：rgznai100） 黄教主的演讲 感觉才没过几天，今年的 GTC 英伟达大会也即将迎来尾声了。 而今年比尔·达利则是对话"AI 教父" 杨立昆 （Yann LeCun），很有前后呼应的感觉。 但 GTC 并不只有黄仁勋和杨立昆，还有许多精彩的演讲与对话，比方说： ………… 接下来的一段时间， CSDN AI 科技大本营 将会在「 GTC 2025 大师谈 」栏目持续更新这些精华内容的全文整理，尽情期待。 比尔·达利 自己就在采访杨立昆之后进行了一场 演讲 ，系统性地讲解了英伟达 2024 一整年的四大项目进展，内容干货很多； OpenAI o1 作者 诺姆·布朗 （Noam Brown）和英伟达的 AI 科学家来了一场 对话 ，他认为现在 AI 圈最需要来一场革命的，就是这些五花八 门的 基准测试 （Benchmark），而且改这个东西还不需要花太多算力资源； 2018 年诺贝尔化学奖得主 弗朗西斯·阿诺德 （Frances Arnold）围绕 AI for Sciense 还有蛋白质工程进行了一场相当硬核的 圆桌对话 ； UC 伯克利教授 彼得·阿比尔 （P ...

Core Insights - The essence of AI has evolved from basic image recognition to a revolution in unsupervised learning, indicating a significant leap in capabilities and understanding [3][4][6] - The acceleration of AI performance is governed by a law of accelerating returns, with capabilities doubling approximately every 5.9 months, leading to exponential growth in intelligence [3][46] - The emergence of AI technologies like ChatGPT marks a pivotal moment in public awareness and interaction with AI, akin to the introduction of the Netscape browser for the internet [10][11] AI Development Milestones - The first major realization of AI's potential occurred around 2007 with Google's advancements, particularly highlighted by the "cat paper" which demonstrated unsupervised learning [3][4] - A second significant moment was in 2016, when breakthroughs in reinforcement learning and deep learning led to revolutionary training methods for machines, exemplified by AlphaGo's success [11][13] - The concept of AI as a tool for enhancing human intelligence is emphasized, with the potential for individuals to significantly increase their cognitive capabilities through effective use of AI [46][48] Skills Required in the AI Era - Three essential skills for thriving in the AI era are identified: mastering AI as a tool, engaging in truth-seeking debates, and fostering human connections [46][49] - The importance of human connection is highlighted, as businesses that prioritize genuine human interaction will likely outperform those relying solely on AI [49][50] Ethical and Philosophical Considerations - The discussion touches on the ethical implications of AI development, emphasizing that the true challenge lies not in the technology itself but in the values and motivations driving its evolution [38][40] - The potential for AI to surpass human intelligence raises questions about decision-making authority and the implications of transferring critical decisions to AI systems [42][43] Future Outlook - Predictions suggest that Artificial General Intelligence (AGI) could emerge as early as 2025, with profound implications for society and human interaction with technology [38][41] - The narrative warns against the dangers of a singular focus on AI's capabilities without addressing the underlying human values that shape its development and application [40][41]

Group 1 - The core viewpoint of the article is that AI is expected to create a market space of up to $10 trillion by potentially replacing 20% of human labor and reducing costs by 20% [2][9] - Key industries that will benefit from the AI revolution include e-commerce, cloud computing, advertising, and financial analysis [2][10] - The company aims to adopt a startup mentality to enhance agility and responsiveness in a competitive e-commerce landscape [3][4] Group 2 - The company emphasizes the importance of empowering younger management team members to make decisions and learn from mistakes [5][7] - The AI strategy is closely linked to the company's cloud computing business, which will benefit from AI deployment across various use cases [10][11] - The pursuit of Artificial General Intelligence (AGI) is seen as a philosophical challenge, with implications for understanding human intelligence limits [12][19] Group 3 - The article discusses the diminishing value of developing the smartest AI in isolation, emphasizing the need for practical applications in real-world scenarios [20][21] - Open-source models are highlighted as a means to democratize AI capabilities, allowing a wider range of developers to innovate [20][21] - The company believes that AI will enhance rather than replace human jobs, improving the quality of work in fields like equity research and law [22]