水皮杂谈 一家之言 兼听则明 偏听则暗 乘势而上 盘面消息 A股三大指数今日集体走强，截止收盘，沪指涨0.87%，收报3870.02点；深证成指涨 1.53%，收报12777.31点；创业板指涨1.77%，收报2980.93点。 沪深两市成交额达到 18121亿，较昨日放量844亿。 老水看盘 美国推出人工智能领域的 " 创世纪计划 " ，受该消息刺激，美股三大指数大幅上扬。其中纳 斯达克指数上涨 2.69% ，不仅收复上周五失地，还实现逾三百点的涨幅。 此次领涨的并非英伟达，而是谷歌，其盘中涨幅一度达到 6% 左右。谷歌之所以能独树一帜 ，替代英伟达的领袖地位 ，一方面得益于其推出的机密大模型，另一方面该大模型未采用英伟 达的 GPU ，而是搭载自研 TPU ， 独辟蹊径， 走出了差异化路线。 这一现象也表明，人工智能领域的发展路径呈现多元化特征，当前美股市场已形成多款大模型 齐头并进、硬件设施路径多元的竞争格局。这对我国相关领域发展颇具启发 —— 后发未必无 优势，国内无论是 DeepSeek 、千问等大模型，还是华为推出的算力优化方案，均有望实现 弯道超车或应用场景突破。 总体来看，我国在人工智能应用领 ...

如今，大模型蓬勃发展，有没有指标来衡量AI大模型的"智力能力水平"？近日，清华大学研究团队提出 了大模型的密度法则，相关论文刊发于自然子刊《自然·机器智能》上。密度法则揭示大语言模型的最 大能力密度随时间呈指数级增长，2023年2月至2025年4月，约每3.5个月翻一倍。 计算机领域的"摩尔定律"大家已经耳熟能详：芯片上可容纳的晶体管数量，每隔一段时间就会翻一番。 计算机的强大，不是因为芯片变得像房子一样大，而是因为它在指甲盖大小的面积上集成了天文数字的 计算单元。清华大学计算机科学与技术系助理研究员肖朝军告诉科技日报记者，大模型的智力能力水平 应该也有一个指标，这就是"能力密度"。 研究的核心假设是，采用相同制造工艺、充分训练的不同尺寸模型，其能力密度相同。正如芯片行业通 过提升电路密度实现了计算设备的小型化和普惠化，大模型也在通过提升能力密度实现高效化发展。 肖朝军说，过去，在规模法则的指引下，大家关心一个大模型的"块头"（参数量）有多大，越大就越智 能，就像关心一个举重运动员的体重，体重越重，力量越大。现在，密度法则从另一个视角揭示了大模 型"高效发展"的规律——我们更应该关心它的"能力密度"，即每一单 ...

上海黄浦江畔，杨浦向新而行，昔日工业锈带已蜕变为创新秀带。在这里，无人机划过长空配送货物， 人工智能赋能千行百业，高校实验室的原始创新成果正加速转化为现实生产力。今年3月，《杨"数"浦 新质秀带创新区建设方案》正式发布，提出4方面23项重点任务，统筹推进一揽子改革举措，激发全域 创新内生动力。全域创新，意味着创新要素在杨浦的每一寸土地上自由流动、深度融合，杨浦将聚全域 资源、举全区之力，让创新在每个街角、每个园区、每个课堂自由绽放。 集聚优势打造科创策源新高地 在复旦大学江湾校区附近的湾谷科技园，复铼智能科技有限公司的实验室里，研发团队正攻克大模型产 业化的一道难关。湾谷科技园与复旦大学江湾校区仅一街之隔，依托这一地理优势，复铼团队与高校实 验室保持紧密合作，力争将深耕多年的大模型研究转化为市场竞争力。 杨浦区充分发挥区域高校林立、人才集聚的资源禀赋优势，通过制定《上海市科学技术奖杨浦区奖励实 施办法》等措施，有效调动了科技工作者的积极性。今年1—6月，区域高校在CNS三大国际期刊发表论 文26篇，充分展现了区域原始创新力的提升。 杨浦区搭建多方交流平台，联动"大学""大厂"创新资源，以"三个一"形式（一个高 ...

Core Insights - Several banks, including Agricultural Bank of China, Minsheng Bank, Bank of China, and others, have initiated recruitment for postdoctoral researchers for 2026, focusing on strategic research and innovation in the financial sector [1][3][4] - The recruitment emphasizes interdisciplinary backgrounds, particularly in artificial intelligence, big models, and financial technology, reflecting a trend towards integrating technology with traditional finance [5][6] Recruitment Conditions - The recruitment criteria have become stricter, with banks favoring candidates with interdisciplinary and practical financial experience [3][4] - Specific requirements include a PhD obtained within the last three years or expected graduation by July 2026, with an age limit of 35 years [3][4] - Preferred fields of study include economics, finance, statistics, applied mathematics, computer science, big data, and artificial intelligence, with a strong emphasis on research capability and dedication [3][4] Research Focus - The research topics set by banks highlight their strategic priorities, with a strong focus on AI, big models, and financial technology applications [6][7] - Examples of research topics include "AI Empowering Digital Transformation of Commercial Banks" and "Application of AI and Big Model Technology in Risk Management," indicating a direct response to industry needs [6][7] Industry Trends - The recruitment surge is driven by the dual pressures of transformation in the banking sector and technological opportunities, with banks seeking to enhance core competitiveness through technology [7][8] - AI technologies are seen as crucial for improving operational efficiency and risk management, which are key demands in the current banking landscape [7][8] - The focus on practical research outcomes distinguishes bank postdoctoral programs from academic institutions, emphasizing the need for research that addresses real business challenges [7][8]

1.当前美国算力投资是在通用人工智能越来越近的趋势下，所做的超前基础设施布局。总体上，当前美国算力基建已有泡沫迹象，但尚未到失 控的边缘，虽不能完全排除算力泡沫破裂可能性，但无论从当前大模型技术的快速迭代，大模型企业高速增长的营收，以及各行业上云落地的 需求，都让这一巨额投资具备一定的合理性。 2.美国当前规划建设的大型数据中心项目总装机容量已突破 45 吉瓦（GW），这场建设热潮预计将吸引超 2.5 万亿美元投资。大西洋月刊文章 设想了一种"崩盘机制"，即当这些巨额投资未能按预期变现、资本市场情绪转向、防御性抛售开始，就可能触发系统性的回落，科技股回落→ 估值压缩→投资放缓→AI 基础设施、芯片、数据中心等领域连锁受累。 3.大模型企业的增长斜率足够高，对芯片供应的需求持续提升。从收入来看，以OpenAI和Anthropic为代表的美国企业在个人用户和企业用户 两端均已实现可观收入。如预计今年底OpenAI的年化收入将超过200亿美元，比之前预测的130亿美元大幅增长，相比去年的40亿美元更是增 长5倍，并计划到2030年增长至数千亿美元。 4.今年三季度，亚马逊、微软和谷歌的云计算收入受AI拉动，分别达3 ...

帅师指出，上海金融业联合会作为连接政府、金融机构、科技企业的桥梁，积极发挥平台的作用，推动人工智能在金融领域的应用与创新：一是成立了 金融科技专业委员会，搭建跨业态金融机构间的交流合作平台，促进人工智能等数字技术在金融领域的应用与推广；二是积极推进普惠金融顾问制度，引导 金融机构加大力度服务科技型企业，进一步促进创新成果转换，推动前沿科技在金融领域的落地和发展；三是协同推动区域数字金融高质量发展，联合会积 极发挥长三角金融业发展联盟的作用，与同业共谋长三角乃至全国金融科技高质量发展的新路径。 华东师范大学党委副书记孟钟捷表示，本次论坛以"AI FOR ALL"为主题，聚焦新科技时代的智能金融应用，意义深远。我们充分发挥经济数据科学、 计算机、人工智能等多学科交叉优势，积极布局金融科技前沿，通过成立长三角金融科技研究院，创设上海人工智能金融学院等，致力于构建"人工智能 +金融"的协同创新平台和人才培养高地。 长三角金融科技创新与应用全球大赛组委会主席、华东师范大学长三角金融科技研究院院长陈琦伟认为，长三角作为中国经济最发达的地区在新时期要 深化价值的变现，这是独特的任务和机会，在这个过程当中人工智能起到独特的作 ...

大批新研发的机器人出现在2025世界计算大会创新成果展示区。新华社记者 苏晓洲 摄 大模型、机器人、脑机接口……正在长沙举行的2025世界计算大会上，从主题报告、专题活动到创新成 果展，很多内容展现算力新进展，让人感受到"数字经济"的时代脉动。 本届大会创新成果展示，呈现出很多独特"风景"。 大模型是主展区各家参展单位的"标配"。有参展单位说，随着算力提升，多款千亿参数开源大模型相继 问世。中小企业无需巨额投入，可以直接在开源模型基础上开展微调和二次开发。企业和个人使用开源 大模型研发信息技术应用产品，如今变得日益普遍。 在展会现场，机器人"才艺"闪耀，令人目不暇接。大会创新成果展示区，有机器人乐队演奏、机器"艺 术家"作画、物流机器人"搬砖"、文旅机器人唱歌、装配机器人"打螺丝"，还有医疗、康养、陪护、运 动助力等各类机器人的发布、演示、售卖…… 大会开幕式上，主办方发布了2025全球计算十大创新成就。其中"全球算力水平首次达到ZFLOPS级"位 居首位。专家介绍，日益成熟的ZFLOPS级计算支撑了超百万款计算服务的商业化落地，覆盖金融、制 造、医疗等关键领域，还催生了"行业大模型"等一系列新业态，在激活数 ...

Core Viewpoint - The rapid advancement of generative AI and large models contrasts with the slow commercial adoption, as evidenced by a recent decline in the percentage of U.S. companies using paid AI products [2][3]. Group 1: AI Project Challenges - Approximately 90% of vertical enterprises do not truly understand AI, leading to ineffective implementation without tailored models [3]. - The telecom industry has historically absorbed new technologies, and AI is seen as the next evolution, with significant advancements expected by 2025 [3]. Group 2: Future of AI and Agent Technology - The AI industry is at a crossroads, with a shift from foundational model development to large-scale application deployment, raising questions about the future of basic model research [6]. - There is a consensus that future AGI will rely on world models that integrate multiple modalities, although specific applications may require tailored models for efficiency [6][7]. - The development of specialized models for various industries is viewed as a practical approach to achieving commercial viability before moving towards universal models [7]. Group 3: Agent Technology Implementation - By 2025, agent technology is expected to become a core trend, with practical applications emerging across various industries, including healthcare and education [8]. - Current implementations of agent technology have demonstrated effectiveness, with plans for broader deployment in 2026 [8]. - Challenges remain in integrating agents into existing workflows, primarily due to limitations in multi-modal capabilities of large models [8][9]. Group 4: Computational Power and Industry Growth - The AI industry faces ongoing challenges related to computational power, with domestic GPU companies accelerating their development to address these needs [9]. - As computational issues are resolved, significant advancements in multi-modal models and agent technology are anticipated [9][10]. Group 5: Consumer Acceptance and Market Trends - Consumer acceptance of AI products is increasing, with a shift towards deploying AI capabilities from cloud to edge devices [9][10]. - The mobile AI sector is expected to see rapid growth, with small models achieving high accuracy in practical applications [11]. Group 6: Humanoid Robots and Industry Development - Humanoid robots are still in the exploratory phase, with significant technical challenges remaining before widespread commercial deployment [12][13]. - The manufacturing of humanoid robots involves complex components, with a focus on developing autonomous control capabilities as a critical bottleneck [13]. - The path to commercial viability for humanoid robots is expected to begin in industrial settings before expanding to consumer applications [14][15].

人形机器人是一类在外观结构和运动方式上尽量接近人类、能够模仿人类行为的机器人，通常具有与人 类相似的躯干、四肢等身体结构。2025年5月，全球首个《人形机器人智能化分级标准》团体标准正式 发布，为人形机器人智能化能力的分级评估、技术研发和应用推广提供了统一的技术语言和评价体系。 2025年度十大科普热词发布 大模型、人形机器人、智能体等入选 中新网北京11月21日电 (记者 孙自法)记者11月21日从中国科普作家协会获悉，在当日举行的2025年全 国科普创作大会上，中国科普作家协会发布2025年度十大科普热词，大模型、人形机器人、智能体、科 幻产业等入选。 2025年度十大科普热词具体包括：全国科普月、科学家精神、大模型、低空经济、人形机器人、智能 体、创新文化、工业遗产、场景创新、科幻产业。它们分别从科技、文化、社会等维度，综合勾勒出 2025年中国科普事业发展、科技前沿动态、科学传播与社会文化融合的整体态势和核心方向，是当下以 及接下来一段时间科普创作者应该重点关注的方向和领域。 其中，大模型是一类多基于深度神经网络构建、具有海量参数的人工智能模型，包括大语言模型、视觉 大模型、多模态大模型以及面向科研的 ...

Core Points - The "Zhongguancun AI Enterprises Overseas Service Port" was officially launched at the "2025 AI+ Conference," aiming to assist domestic AI companies in global market expansion [1][3] - The service system includes "Overseas Service Port" and "Overseas Service Station," providing comprehensive support for companies before and after their international ventures [2][3] Group 1 - The four initial organizations, including Zhongguancun Jinggangao Youth Innovation Center and Zhongguancun Development Group, were selected for their unique capabilities in facilitating overseas expansion for AI companies [4] - The service system has already successfully assisted companies like Ruile Intelligent, Yunjike Technology, and Zhongke Wenge in establishing a presence in Hong Kong's Cyberport [4] - The establishment of the service port and station has received positive feedback from AI enterprises, highlighting its strategic importance in enhancing efficiency and competitiveness in international markets [5] Group 2 - Beijing has positioned Guangxi as a "bridgehead" for expanding into ASEAN markets, facilitating the overseas deployment of AI technologies and products [5] - The service port and station address critical pain points such as trust establishment, compliance risks, and resource connectivity, acting as a "trust bridge" and "accelerator" for companies [5] - The combined service system is expected to significantly lower the barriers for small and medium-sized AI enterprises looking to expand internationally [5]