来源：环球网 此次全员大会披露，xAI 正进行组织架构升级调整，伴随架构调整出现了部分员工离职的情况，其中创 始团队也有不少成员离开。马斯克对此表示，快速发展的企业需要适配的组织架构，此次调整虽导致部 分员工离开，但公司也在积极开展招聘工作，尤其面向月球大规模自动驾驶汽车项目吸纳人才。 据介绍，xAI 新的组织架构划分出四大核心团队，各有明确聚焦方向：团队一专注于含语音功能的 Grok 聊天机器人研发，团队二负责应用编码系统搭建，团队三主攻 Imagine 视频生成工具优化，团队 四则聚焦 Macrohard 项目研发。该项目覆盖从简单计算机操作模拟到全企业建模的广泛领域，负责该项 目的托比・波伦表示，Macrohard 技术能够完成电脑可执行的各类任务，未来人工智能甚至有望完整设 计火箭发动机。 大会还公布了 xAI 与 X 平台的最新运营数据，亮眼的成绩展现了相关产品的市场表现。X 平台产品负 责人尼基塔・比尔透露，X 的订阅业务年经常性收入已突破 10 亿美元，这一成果得益于节假日期间的 营销推广。xAI 方面则披露，旗下 Imagine 工具日均生成视频量达 5000 万条，过去 30 天生成图片数量 ...

Group 1 - Apple has completed the acquisition of Canadian graph database company Kuzu in October 2025, marking a significant move in the data technology sector [1] - Kuzu, founded in 2023 and based in Ontario, Canada, had around 10 employees at the time of acquisition, and its website has since been taken down [1][4] - The acquisition is seen as a strategic shift for Apple, which has traditionally relied on relational databases, as Kuzu specializes in graph databases that offer efficiency advantages in multi-layered queries [4] Group 2 - The technology from Kuzu may have multiple future applications, including potential integration into Apple's iWork suite to enhance collaborative features [5] - The visual characteristics of graph databases could align with mind-mapping tools, suggesting possible enhancements to Apple's productivity applications [5] - There is speculation that the technology could improve social and content-sharing features across Apple's platforms, such as Game Center and Apple Music [5]

该类器官模型能够重现损伤后的典型变化，包括细胞死亡、炎症反应以及胶质瘢痕形成。胶质瘢痕是由 密集疤痕组织构成的物理化学屏障，长期以来被认为是阻碍神经再生的主要障碍。 团队重点测试了一种被称为"跳动分子"的新型疗法。该疗法在早前动物实验中已显示出能促进组织修复 并逆转瘫痪的效果。在本次类器官实验中，受损组织在治疗后表现出显著的神经元轴突生长，同时胶质 瘢痕样组织也明显减少。这些结果进一步支持该疗法具有转化为临床应用的潜力，其已获得美国食品药 品监督管理局的认证，有望为脊髓损伤患者提供新的治疗选择。 来源：科技日报 科技日报北京2月11日电 （记者张梦然）美国西北大学科学家开发出迄今最先进的人类脊髓损伤类器官 模型，能精准模拟脊髓损伤的关键病理特征，并为测试新型再生疗法提供了高效平台。该研究首次利用 实验室培养的人类脊髓类器官（即由干细胞衍生的微型器官结构）模拟了不同类型脊髓损伤，并验证了 具有潜力的新型"跳动分子"治疗策略。相关成果发表于新一期《自然·生物医学工程》。 该脊髓类器官直径可达数毫米，具备较高的成熟度和复杂性，包含神经元、星形胶质细胞以及首次被引 入的中枢神经系统免疫细胞。这使得模型能更真实地模拟损 ...

Core Viewpoint - The article highlights the launch of a new promotional campaign by Migu Yuedu (formerly Migu Reading) to enhance user engagement during the Spring Festival, offering a 30-day free membership experience to access a wide range of digital content [1][11]. Group 1: Strategic Upgrade - Migu Yuedu has undergone a strategic upgrade, officially rebranding from Migu Reading to Migu Yuedu on January 29, 2026, expanding its content offerings to include both literature and short dramas [3]. - The platform aims to create a new integrated experience by combining reading and viewing, focusing on user satisfaction and immersive content consumption [3][13]. Group 2: User Experience Enhancements - The upgraded platform provides users with 30GB of dedicated data each month, addressing concerns about data usage while streaming content [6]. - Migu Yuedu features a dual-channel system, "Book City" and "Theater," which includes over 16,000 micro-short dramas and a vast collection of licensed literature, covering popular genres [6]. - The platform employs intelligent recommendation algorithms to seamlessly integrate literature and short drama content, enhancing user experience [6][8]. Group 3: Target Audience and Inclusivity - Migu Yuedu has implemented features catering to various age groups, including an "Elderly Mode" with larger fonts and simplified navigation for older users, as well as special content for students [10]. - The platform's initiatives reflect a commitment to bridging the digital divide and addressing the cultural needs of diverse user demographics [10][14]. Group 4: Cultural Engagement and Long-term Strategy - The promotional activities align with national reading initiatives, aiming to cultivate digital content consumption habits among the public [11][14]. - Migu Yuedu seeks to establish a complete ecosystem that connects literature and visual content, reinforcing its competitive edge in the micro-short drama and digital reading sectors [13]. - The company is dedicated to contributing to cultural dissemination and industry upgrades, positioning itself as a leader in the digital cultural landscape [13][14].

Core Insights - The development of practical fault-tolerant quantum computers, once thought to be decades away, is accelerating, with expectations for a million-qubit scale machine by 2035 [2] Group 1: Advances in Quantum Computing - Multiple leading research teams have made significant progress in quantum error correction, gate fidelity, and system integration, indicating that large-scale fault-tolerant quantum computing is physically achievable [3] - Four independent research teams have confirmed that their quantum systems have surpassed the critical error rate threshold necessary for fault-tolerant computing [3] Group 2: Error Correction Efficiency - Current estimates suggest that one logical qubit requires thousands of physical qubits, with a redundancy ratio of "1000:1," which exceeds current engineering capabilities [5] - Innovations in algorithms and architectures have led to a significant reduction in resource consumption, with a new method proposed by Google reducing the number of qubits needed for large integer factorization from 20 million to 1 million, a decrease of two orders of magnitude [5] Group 3: Gate Fidelity Improvements - Research teams have achieved remarkable breakthroughs in gate fidelity, with Oxford University reporting a single qubit gate fidelity of 99.999985%, a tenfold improvement over previous records [6] Group 4: Extending Coherence Time - Enhancements in qubit coherence time are crucial for improving hardware durability, with Princeton University extending superconducting qubit lifetimes from 0.1 milliseconds to 1.68 milliseconds [7] - If coherence times can reach 10 milliseconds, the resource overhead for error correction could be reduced by two to three times, making large integer factorization tasks feasible with 30,000 to 50,000 qubits [7]

Core Insights - The global intelligent robotics company Geek+ has launched its first general-purpose robot, Gino 1, designed specifically for warehouse scenarios [1] - Gino 1 is trained for multi-tasking capabilities, enabling it to perform various warehouse operations such as picking, box moving, packaging, and inspection [1] - The robot's hardware is tailored to address warehouse pain points, ensuring accuracy and safety through advanced visual systems and dexterous manipulation [1] Company Developments - Gino 1 integrates Geek+ Brain, which combines extensive warehouse data with large-scale simulation reinforcement learning, providing human-like operational capabilities [1] - The robot is designed to meet user demands in terms of reliability, efficiency, and cost-effectiveness, facilitating its commercial application [1] - Gino 1 has achieved mature mass production capabilities, allowing for large-scale deployment [1] Industry Context - Over 70% of warehouses globally still rely heavily on manual labor, indicating a significant opportunity for automation solutions [2] - Geek+ aims to accelerate the implementation of unmanned warehouse solutions through a closed-loop iteration of "scenario-data-technology," promoting both technological and commercial evolution [2]

Core Insights - The recent launch of "LobsterAI" by NetEase Youdao marks a significant advancement in personal assistant agents, showcasing capabilities that extend beyond simple interactions to complex task execution [1][2] Group 1: Product Features - "LobsterAI" integrates innovative features, combining autonomous cross-application task execution similar to "OpenClaw" with a graphical user interface akin to "Claude Cowork" [2] - The product aims to enhance user experience by offering long-term memory and scheduled task functionalities, allowing for a more personalized interaction [2] - Users can interact with "LobsterAI" through mobile and PC platforms, enabling remote engagement via applications like DingTalk and Feishu [2] Group 2: Functionality and Use Cases - "LobsterAI" can perform various tasks such as information retrieval, schedule management, and in-depth data analysis through conversational commands [2] - The agent supports a timed task mechanism, allowing users to automate daily tasks like collecting industry news [2] - It possesses long-context memory capabilities, enabling it to understand user preferences over multiple interactions for a more efficient experience [2] Group 3: Security Measures - To address concerns regarding automated operations, "LobsterAI" employs a strict "local-first" strategy, operating within a sandbox environment to prevent system file damage [3] - The system supports local data processing to mitigate risks of cloud data leakage [3]

团队将目光投向一类特殊的神经元——"记忆印痕细胞"。这些神经元在学习时被激活，在回忆时被再 度"点亮"，构成大脑中真实的"记忆痕迹"。但在老年动物或阿尔茨海默病模型动物中，这些印痕细胞常 出现功能紊乱，导致记忆提取失败。若能在衰退发生后，重新激活这些记忆印痕神经元，就可能找回丢 失的记忆。 团队采用部分重编程技术，短暂启动三个关键基因Oct4、Sox2和Klf4（简称OSK）。此前研究表明，这 组因子可在一定程度上逆转细胞老化迹象。但最新研究创新之处在不泛化作用于全脑，而是精准锁定记 忆印痕神经元。 他们利用腺相关病毒作为载体，通过精确脑部注射，将标记学习激活神经元的荧光系统和可控开启OSK 表达的时间开关两个元件送入两大关键脑区——海马体齿状回与内侧前额叶皮层。前者主导近期记忆的 形成与提取，后者则负责远期记忆和回忆。 来源：科技日报 科技日报记者 刘霞 瑞士洛桑联邦理工学院脑心智研究所科学家在10日出版的《神经元》杂志上发表论文指出：通过重编程 与记忆相关的特定神经元，可有效恢复多种疾病模型小鼠的记忆功能。 结果显示，在老年小鼠中，仅需短暂激活海马体内的印痕神经元OSK表达，其记忆表现便恢复至年轻小 鼠水 ...

Core Viewpoint - The successful integration of the first domestically produced voltage-source static frequency converter (SFC) at the Nanning pumped storage power station marks a significant advancement in China's independent research and development of new pumped storage technology, enhancing the safety of the energy storage supply chain and promoting the construction of a new power system [1][4]. Group 1: Technological Advancements - The new voltage-source SFC significantly reduces harmonic distortion in output current, achieving a total harmonic distortion rate of only 0.9%, well below the design requirement of 3% [6]. - The acceleration performance of the new SFC improves by over 35% compared to the previous current-source SFC, enhancing the ability to accelerate the motor from a standstill to rated speed [6]. Group 2: Industry Context and Future Prospects - With the rapid development of renewable energy in China, the demand for high-performance control devices in pumped storage systems is increasing, necessitating advancements in technology [4]. - The long-term development plan for pumped storage in China includes over 580 key and reserve projects, with a total capacity exceeding 720 million kilowatts, indicating a broad market for the new SFC [7]. - The company plans to accelerate the pilot application of the new SFC in more pumped storage units and expand its application in areas such as energy storage, phase adjustment, and gas turbines, promoting deeper integration of technological and industrial innovation [7].

Core Insights - Venus and Earth are similar in size and origin but have vastly different environments, with Venus having surface temperatures exceeding 480°C and atmospheric pressure nearly 100 times that of Earth [2] - Radar data from NASA's Magellan spacecraft revealed that Venus's surface is dominated by volcanic features, indicating a geological evolution heavily influenced by volcanic activity [2] Group 1 - The research focuses on the presence of collapse pits or "skylights" on Venus's surface, which may indicate the existence of underground cavities [3] - A significant collapse pit was discovered in the Nix Mountains region, suggesting a possible connection to a large underground channel with a diameter of approximately 1 kilometer and a depth exceeding 375 meters [3] - The unique environmental conditions on Venus, such as lower surface gravity and denser atmosphere, may allow for the formation of larger underground cavities compared to Earth [3] Group 2 - Although the findings are promising, the evidence remains indirect, and confirmation of these channels requires higher resolution radar and new exploration missions [3] - Future missions, including ESA's "EnVision" and NASA's "Veritas," are expected to provide advanced radar and imaging capabilities to further investigate Venus's surface and subsurface structures [3]