那种让读者在关掉无数窗口后，依然愿意停下来，想一想 "哦，事情原来是这样的" 的内容。 我们希望可以提供可靠的信息、知识与判断，帮助人们理解世界的复杂与变化；努力提出一些真问 题，也让一些被忽视的声音被更多人听到。 以及，让当代中国企业家最好的访谈都发生在这儿。 晚点的价值主张是什么？ 晚一点、好一点 晚点是谁？ 一家还在努力基于一手信息写稿的公司。成立于 2019 年 ，生日是 4 月 1 号，团队规模 49 人，平均 年龄 31.8 岁。过去 6 年，我们记录了中国最重要的公司、人物与商业事件，但我们知道，这仅仅是开 始。 晚点想成为什么？ 中国最好的商业媒体。 不仅关注科技与 AGI，也是充满善意的，关心人类与社会演进的媒体。 我们想给用户提供怎样的内容？ 强调专业主义；相信文字依然能承载人类的好奇心和理性。 我们想招募怎样的同事？ 这里没有公式化的报道，有新闻工作经验很好，但不是必需。我们希望招募对商业、科技、社会变迁 有深刻理解的写作者；擅长讲述故事、擅长访谈，或是对数字和数据敏感，能拆解复杂的系统。 具体岗位： 工作地点： 北京、上海、广州、深圳 我们最看重的是： 行业里有人会谈论，晚点的编辑太严 ...

在大多数人的直觉里，只要经济持续增长，工资总会水涨船高。 可最新的一篇论文却抛出残酷预言：在AGI时代，经济会因为算力扩张而狂飙不止，但普通人的工资却被「算力成本」锁死，彻底与增长脱钩。 也就是说，哪怕GDP翻十倍，你的收入可能一分钱没涨。 真正的问题不只是失业，而是——当劳动不再驱动增长，我们的价值由谁来定价？ AGI时代或将带来前所未有的繁荣：算力推动经济狂飙，但人类工资却被钉死在「算力成本」上，与增长彻底脱钩。耶鲁学者Restrepo的研究 指出，劳动份额将归零，财富全面流向算力资本。人类或许仍被需要，却只停留在护理、陪伴等附属岗位。在这样的未来，工作还有意义吗？ 论文链接：https://conference.nber.org/conf_papers/f227505.pdf 当工资被算力定价，人类能靠什么涨薪？ 在传统经济学里，工资往往与「技能稀缺性」挂钩：你能完成别人做不了的事，就能拿到溢价。 但Restrepo在论文里抛出一个惊悚结论：在AGI经济中，工资不再取决于你的技能有多稀缺，而是取决于复现你这份技能所需的算力成本。 论文里有一个关键公式： $\chi_{t}(a)=L_{t}(a)+\fr ...

Over at the UN, more warnings about potentially apocalyptic civilization ending scenarios if AI doesn't get some strong guard rails soon. This time we are talking more than scientists. We are talking Nobel Prize winners, even two of the three so-called godfathers of AI are raising the alarms, urging political leaders across the world to put some meaningful regulation up by the end of next year.And here's what one Nobel Peace Prize winner said earlier today on the floor. We urge your governments to establish ...

AGIX 指数诞生于我们对"如何捕获 AGI 时代 beta 和 alphas"这一问题的深度思考。毫无疑问，AGI 代表了未来 20 年最重要的科技范式转换，会像互联网那样重塑了人类社会的运行方式，我们希望 AGIX 成为衡量这一新科技范式的重要指标，如同 Nasdaq100 之于互联网时代。 「AGIX PM Notes」 是我们对 AGI 进程的思考记录，希望通过学习 Warren Buffett、Ray Dalio、 Howard Marks 等传奇投资者们的分享精神，与所有 AGIX builders 一同见证并参与这场史无前例的技 术革命。 | Ticker | 本周表现 | YTD | Return since 2024 | | --- | --- | --- | --- | | AGIX | 3.11% | 31.66% | 92.48% | | S&P 500 | 0.74% | 13.31% | 39.72% | | QQQ | 1.30% | 17.24% | 46.35% | | DowJones | 0.94% | 8.86% | 22.89% | | | 本周表现 | Index ...

据报道，2025阿里云栖大会将于9月24日至26日在浙江杭州云栖小镇举行。 消息面上，北京时间9月18日凌晨，美联储联邦公开市场委员会（FOMC）会议纪要释放关键政策信号——将联邦基金利率目标区间下调25个基点至 4.00%-4.25%。这是美联储今年首次降息，也是继2024年三次降息后再次降息。 截至2025年9月22日 10:39，恒生科技指数下跌0.83%。成分股方面涨跌互现，舜宇光学科技领涨4.54%，中芯国际上涨4.16%，百度集团-SW上涨2.97%；同 程旅行领跌3.29%，美团-W下跌3.29%，快手-W下跌3.10%。恒生科技指数ETF(159742)下跌0.12%，最新报价0.85元。拉长时间看，截至2025年9月19日，恒 生科技指数ETF近1周累计上涨4.79%，涨幅排名可比基金1/10。 流动性方面，恒生科技指数ETF盘中换手7.78%，成交3.22亿元。拉长时间看，截至9月19日，恒生科技指数ETF近1周日均成交8.96亿元。 资金方面，截至9月19日，南向资金上周累计净流入额达368.51亿港元，已连续18周净流入。 份额方面，恒生科技指数ETF最新份额达48.48亿份，创近1年 ...

Core Insights - The AI Creator Carnival held in Haidian District showcased a collaborative innovation ecosystem, attracting over 10,000 participants including developers, entrepreneurs, and students [1][2] - The event featured over 40 thematic sub-venues and 80 interactive exhibition booths, highlighting the integration of AI across various sectors such as entertainment, education, culture, and finance [1][3] Group 1 - The carnival was guided by the Zhongguancun Science City Management Committee and co-hosted by Silicon Star, Zhongguancun Science City Company, and Zhongguancun Entrepreneurship Street [1] - More than 100 AI ecosystem partners participated, breaking traditional barriers and promoting cross-domain collaboration [1][2] - The event emphasized the transition of AI technology from concept to practical application, showcasing the vitality of the AI industry in Haidian District [1][3] Group 2 - The event included over 20 content co-creation units engaging in workshops, talks, and demo days, fostering a vibrant atmosphere for innovation [2] - The main stage featured discussions on future AI entrepreneurship trends, humanoid robots, and 3D gaming technology, encouraging open dialogue among participants [2] - The AGI Bar served as a nighttime hub for idea exchange, while the Zhongguancun Science and Technology Financial Service Center provided financial support to over 30,000 enterprises [2] Group 3 - The AI Technology Market, a core exhibition area, showcased over 80 tech companies and innovative institutions, presenting cutting-edge technologies like no-code development and multi-modal sound synthesis [3] - The event included an "AI Talent Corner" aimed at job seekers, offering direct interactions with over 20 AI companies and providing insights into employment policies [3] - Participants had the opportunity to engage in one-on-one career consultations and discussions with industry professionals, enhancing networking opportunities [3]

Core Viewpoint - Huawei's advancements in AI and computing capabilities, particularly through its Ascend chip series and supernode technology, position it as a strong competitor in the global AI infrastructure landscape, despite facing challenges from U.S. sanctions and competition from companies like NVIDIA [2][3][4]. Group 1: AI and Computing Strategy - Huawei's core strategy revolves around "supernodes + clusters," which aims to provide continuous AI computing power despite limitations in chip manufacturing technology [4][8]. - The company has developed the CloudMatrix 384 supernode, which is the largest commercial supernode in China and boasts the highest effective computing power globally [8][9]. - Huawei plans to release several new AI chips and supernodes over the next few years, with a roadmap that includes the Ascend 950PR, 950DT, 960, and 970 series, aiming for a doubling of computing power approximately every year [12][13]. Group 2: Competitive Positioning - Huawei's Ascend 950 supernode is projected to maintain its status as the world's strongest computing supernode for several years, significantly outperforming NVIDIA's upcoming products in various metrics [13][15]. - The company emphasizes that its supernode architecture is designed to overcome traditional computing limitations, moving away from the von Neumann architecture to a self-defined "equal interconnection architecture" [9][16]. - Huawei's innovations are seen as a response to U.S. sanctions, pushing the company to explore alternative paths for technological advancement [7][8]. Group 3: Ecosystem and Market Dynamics - Huawei is focused on building a robust ecosystem around its hardware, with an emphasis on open-source software to attract developers and enhance hardware sales [17][18]. - The company does not rely on the CUDA ecosystem, instead developing its own MindSpore framework to ensure independence from Western supply chains [18]. - Huawei's long-term vision includes not just chip development but also providing comprehensive system integration capabilities to foster healthy collaborations within the AI industry [18][19].

参考 | 量子位 点击下方 卡片 ，关注" 自动驾驶之心 "公众号 戳我-> 领取 自动驾驶近30个 方向 学习 路线 >>自动驾驶前沿信息获取 → 自动驾驶之心知识星球 本文只做学术分享，如有侵权，联系删文 任少卿去中科大了！ AI大神任少卿开始在母校中国科学技术大学，开班招生了。 任少卿，曾任Momenta联合创始人、蔚来汽车副总裁，07级中科大本硕博（微软亚洲研究院联合培养），ResNet和Faster R-CNN作者。学术高被引超44 万，是全球中国籍学者高被引第一。ResNet也是21世纪全球最高被引论文。获未来科学大奖-数学与计算机科学奖。 招生方向为AGI、世界模型、具身智能、AI4S等。 硕士、博士生都在招。有推免资格的学生，下周一（22日）开启紧急面试。 更多内容 自动驾驶产业和学术最新咨询，欢迎加入自动驾驶之心知识星球，国内最大的自驾社区平台。 ...

Core Viewpoint - Ren Shaoqing, a prominent figure in AI and computer vision, is starting a recruitment program at his alma mater, the University of Science and Technology of China, focusing on advanced topics in AI such as AGI, world models, embodied intelligence, and AI for Science [1][2]. Group 1: Recruitment Details - The recruitment is open for both master's and doctoral students, with emergency interviews starting on the upcoming Monday for students with recommendation qualifications [3]. - Interested students can send their resumes to Ren Shaoqing's email for inquiries regarding the application process and interview details [16]. Group 2: Background of Ren Shaoqing - Ren Shaoqing is an expert in computer vision and autonomous driving, having graduated from the University of Science and Technology of China and obtained a joint PhD with Microsoft Research Asia [4][5]. - He has been recognized as one of the most influential scholars in AI, ranking 10th in the AI 2000 list, and received the Future Science Prize in Mathematics and Computer Science in 2023 [6]. Group 3: Contributions to AI - Ren is a co-author of ResNet, a groundbreaking work in deep learning that addresses the vanishing gradient problem, significantly impacting fields requiring high perception capabilities like computer vision and autonomous driving [7]. - ResNet has received over 290,000 citations and won the Best Paper Award at CVPR 2016 [8]. - He also contributed to Faster R-CNN, an efficient two-stage object detection algorithm that balances speed and accuracy [10]. Group 4: Role in NIO - After completing his PhD, Ren co-founded Momenta and later joined NIO, where he played a key role in developing autonomous driving algorithms and leading the smart driving R&D team [13]. - At NIO, he developed the NIO World Model (NWM), which integrates spatiotemporal cognition and generative capabilities, allowing for high-fidelity scene reconstruction and long-term scenario simulation [14][15].

机器之心报道 机器之心编辑部 科技的走向、股市的涨跌、比赛的胜负、选举的结果…… 在 AI 时代，未来还会像过去那样不可知吗？ 全球首个动态实时 LLM 智能体未来预测基准 FutureX 推 出， 被马斯克称为衡量智能的最佳标准。 9 月 18 日，全球创新企业家、慈善家陈天桥旗下的 MiroMind 团队在这一 AI 未来大考中，连续第二周蝉联冠军。与专注文本输出的生成式模型不同，MiroMind 采用记忆驱动机制，专为预测与决策设计，旨在打造全球最好的预测大模型。 科幻电影早已为我们描绘过可能性：《少数派报告》AI 预测犯罪，让「未发生的事件」变得可防可控；《Her》智能系统能够预测人类行为和选择。今天，这些幻 想正逐步照进现实。 长期以来，AI 评估大多停留在答案已知的静态问题上，如同学生反复练习往年高考试卷。然而，真正有价值的能力是对未知未来的准确预测。投资和规划将不再 依赖盲目试错，可以提前洞察趋势，把握机遇，有效规避风险，真正掌握主动权。毕竟，谁不曾幻想过回到过去，提前买入 Nvidia 或 Tesla 呢？ 正如伊隆・马斯克所说：「预测未来的能力是衡量智能的最佳标准。」让 AI 具备像人类一样 ...