哈工大、鹏城实验室、新加坡国立、复旦、北大 联合发布了一篇重磅综述 《AI Meets Brain: A Unified Survey on Memory System from Cognitive Neuroscience to Autonomous Agents》， 首次打破认知神经科学与人工智能之间的学科壁垒，系统性地将人脑记忆机制与 Agents 记忆统一审视，为设计真正「类人」的 Agent 记 忆系统奠定理论基石。 全文横跨认知神经科学与人工智能两大领域，涉猎相关文献共 400 篇。 跨学科突破：神经科学如何让 Agent 拥有「人类式」记忆？ 你是否想过 Agent 能像人类一样积累经验、不断成长？如今，这一愿景正加速走向现实。但是，现有研究要么只聚焦 AI 技术本身，要么对人脑记忆机制的借鉴浮 于表面，两个学科之间始终缺少真正的灵感碰撞。 什么是记忆？ 综述重新定义了记忆。记忆不仅仅是数据的存储，它也是认知的纽带。综述从认知神经科学到 Agent 对记忆进行了剖析： 1.认知神经科学角度：连接过去与未来的桥梁 在人脑中， 记忆不仅仅是回放信息，其本质是大脑存储和管理信息的过程。 记忆是连接过去 ...

作者 | 允毅、木子 2025 年的硅谷 AI 圈，最激烈的战场已不止于模型参数和榜单上，另一场残酷的战争也在暗中同步升级。 当大模型一路卷到极限，算力、参数规模、基准测试分数开始出现明显的边际递减，真正被重新定价的，是"人"。 过去几年，硅谷 AI 的主叙事是"谁能训练出更大的模型、刷出更高的分数"。 但进入 2025 年，模型能力仍然重要，却不再是唯一的决定因素；大家的关注重心逐渐从"模型参数与评测分数"，转向"谁能够将模型纳入产品与系统核 心，并持续推动其在真实业务场景中发挥作用"。 这一变化，非常直观地体现在一连串人员流动中： 一边是科技巨头高调宣布重金抢人、疯狂扩招 Agent、系统、基础设施方向的研究与工程负责人；另一边，他们又在内部对原有 AI 研究体系进行重组， 让多位中高层研究负责人选择离开舞台中央。 在一系列重大人事变动中，Meta 今年的变化尤为瞩目：比如前两天豪掷 20 亿美元买下智能体公司 Manus，顺手也把 Manus 创始人肖弘"纳入囊中"。另外 据《华尔街日报》7 月报道，Meta 采用"爆炸式 offer"战术：签约金最高达 1 亿美元，决策窗口短至几小时。 而作为 Met ...

除了技术路径的更迭，卡帕西还对智能的本质提出了深刻见解。 在这份综述中，卡 帕西详尽地剖析了过去一年中大语言模型 （LLM） 领域发生的底层范式转移。他指出，2025年标志着AI训练哲学从 单纯的"概率模仿"向"逻辑推理"的决定性跨越。 这一转变的核心动力源于可验证奖励强化学习 （RLVR） 的成熟，它通过数学与代码等客观反馈环境，迫使模型自发生成类似于人类思 维的"推理痕迹"。卡帕西认为，这种长周期的强化学习已经开始蚕食传统的预训练份额，成为提升模型能力的新引擎。 北京时间12月21日，OpenAI创始人之一、AI大神安德烈·卡帕西（Andrej Karpathy）发布了名为《2025年大语言模型年度回顾》（2025 LLM Year in Review）的年度深度观察报告。 他用"召唤幽灵" （Summoning Ghosts） 而非"进化动物" ( E volving/growing Animals) 来比喻当前AI的成长模式，解释了为何当前的大语 言模型会展现出"锯齿状"的性能特征——在尖端领域表现如天才，却在基础常识上可能如孩童般脆弱。 此外，卡帕西也对"氛围编程 （ Vi be Coding） " ...

Core Insights - The article discusses a significant paradigm shift in the field of large language models (LLMs) in 2025, moving from "probabilistic imitation" to "logical reasoning" driven by the maturity of verifiable reward reinforcement learning (RLVR) [2][3] - The author emphasizes that the potential of LLMs has only been explored to less than 10%, indicating vast future development opportunities [3][25] Group 1: Technological Advancements - In 2025, RLVR emerged as the core new phase in training LLMs, allowing models to autonomously generate reasoning traces by training in environments with verifiable rewards [7][8] - The increase in model capabilities in 2025 was primarily due to the exploration and release of the "stock potential" of RLVR, rather than significant changes in model parameter sizes [8][9] - The introduction of the o1 model at the end of 2024 and the o3 model in early 2025 marked a qualitative leap in LLM capabilities [9] Group 2: Nature of Intelligence - The author argues that LLMs should be viewed as "summoned ghosts" rather than "evolving animals," highlighting a fundamental difference in their intelligence compared to biological entities [10][11] - The performance of LLMs exhibits a "sawtooth" characteristic, excelling in advanced fields while struggling with basic common knowledge [12][13] Group 3: New Applications and Interfaces - The emergence of Cursor represents a new application layer for LLMs, focusing on context engineering and optimizing prompt design for specific verticals [15] - The introduction of Claude Code (CC) demonstrated the core capabilities of LLM agents, operating locally on user devices and accessing private data [17][18] - The concept of "atmospheric programming" allows users to create powerful programs using natural language, democratizing programming skills [20][21] Group 4: Future Directions - The article suggests that the future of LLMs will involve a shift towards visual and interactive interfaces, moving beyond text-based interactions [24] - The potential for innovation in the LLM space remains vast, with many ideas yet to be explored, indicating a continuous evolution in the industry [25]

整个2025年，我与无数企业领袖探讨了他们的AI战略，试图了解哪些措施有效，哪些构成了阻碍。随 着时间推移，我注意到有三个趋势在不同公司和行业中反复出现，它们决定了哪些企业能借助AI取得 成功，哪些会陷入困境。现将这些趋势汇总，分享来自AI转型一线的经验教训。 首先，AI在后端任务中的应用正在蓬勃发展，这表明真正能产生实际影响的往往是那些"枯燥"的工作。 第二个趋势与技术无关，而关乎人：企业如何对待员工，对其AI应用的成败至关重要。然而，或许最 能说明问题的趋势是关于初始战略和动机的：追逐AI技术本身的企业往往失败，而以解决问题为出发 点的企业则能取得成功。 当然，成功的因素远不止这些——从数据治理到安全与合规。但上述这些趋势，无论好坏，正在塑造企 业的AI实践。 摒弃"为AI而AI" 咨询公司韦斯特门罗（West Monroe）的AI与新兴技术负责人埃里克·布朗（Erik Brown）今年早些时候 告诉《财富》，他目睹了许多公司在概念验证未能达到预期后，陷入了"AI疲劳"。他指出，陷入此境地 的企业有一个共同点：要么选错了应用场景，要么误解了AI在该任务中可能（或不可能）发挥的作 用。更具体地说，他们的出发 ...

Core Viewpoint - LeCun expresses skepticism about the potential of large language models (LLMs) to achieve artificial general intelligence (AGI), arguing that the path to superintelligence through LLMs is fundamentally flawed [2][78]. Group 1: Departure from Meta - LeCun is leaving Meta after nearly 12 years, criticizing the company's increasingly closed approach to research and its focus on short-term projects [3][11][26]. - He plans to establish a new company named Advanced Machine Intelligence (AMI), which will prioritize open research and focus on world models [10][19]. Group 2: World Models vs. LLMs - LeCun believes that world models, which handle high-dimensional and continuous data, are fundamentally different from LLMs, which excel at discrete text data [28][29]. - He argues that relying solely on text data will never allow AI to reach human intelligence levels, as the complexity of real-world data is far greater than that of text [31][32]. Group 3: Research Philosophy - LeCun emphasizes the importance of open research and publication, stating that without sharing results, research lacks validity [15][17]. - He critiques Meta's shift towards short-term projects, suggesting that true breakthroughs require long-term, open-ended research [18][26]. Group 4: Future of AI - LeCun envisions that the development of world models and planning capabilities could lead to significant advancements in AI, but achieving human-level intelligence will require substantial foundational work and theoretical innovation [84][85]. - He asserts that the most challenging aspect of AI development is not reaching human intelligence but rather achieving the intelligence level of dogs, as this requires a deep understanding of foundational theories [88][89]. Group 5: Personal Mission - At 65, LeCun remains committed to enhancing human intelligence, viewing it as the most scarce resource and a key driver for societal progress [92][94]. - He reflects on his career, expressing a desire to continue contributing to the field and emphasizing the importance of open collaboration in scientific advancement [103].

Core Viewpoint - The future of China is seen as a potential surpassing of the United States, akin to Japan's rise in the 1980s, with the main risks stemming from internal errors rather than external constraints [1][15]. Group 1: Unexpected Developments - The rapid proliferation of smartphones has redefined industry landscapes, integrating multiple functionalities into a single device, which exemplifies a "non-linear explosion" driven by technological convergence [5][6]. - The slow development of VR technology contrasts sharply with the smartphone boom, highlighting the importance of system dependencies and the need for a comprehensive sensory experience for true immersion [8][9]. - The emergence of large language models (LLMs) represents an unexpected breakthrough in AI, showcasing the potential for logic and reasoning through language, diverging from traditional AI paths [11][12]. Group 2: Future Outlook for China - Key drivers for China's growth in the next 25 years include the default option of open-source technology, the confidence of tech innovators moving beyond imitation, and a cultural shift led by returnees embracing globalization [16]. - The exploration spirit that embraces "unexpected beauty" is deemed crucial for fostering innovation and overcoming challenges [17].

（原标题：凯文?凯利：意外之美｜我们的四分之一世纪） 编者按： 2025 年，经济观察报以 " 我们的四分之一世纪 " 为年终特刊主题，旨在通过数十位时代亲历者 的故事，共绘一幅属于这段岁月的集体记忆图谱。 二十一世纪的第一个25年即将结束之际，我问凯文·凯利（KK），有哪些发展超乎他的想象？又有哪些 低于他的预期？他简单地将答案归结为"意外之快""意外之慢"，以及"意外之路"。 这三大意外也让我们意识到，前瞻未来时，我们往往会低估创新者的颠覆性，因此必须跳出线性思维； 我们也会在一厢情愿中忽略木桶原理；当然，也会有意外之喜，因为另辟蹊径常常带来爆炸式的后果。 归根结底，未来既充满不确定性，也蕴藏诸多可能性，发现意外之美是最大的快乐。 一、 意外之快：智能手机的"非线性爆发" "我没想到智能手机会吃掉一切"——智能手机普及的速度与广度——是KK最直观的意外。2007年 iPhone问世时，多数人将其视为更精致的功能机；但短短十年间，它不仅完成了全球数十亿级的渗透， 更以吞噬一切的姿态重构了产业格局：相机、MP3、导航仪、钱包乃至电脑的功能，被压缩进方寸屏 幕；社交媒体、移动支付、网约车等新生态，借由手机的普及 ...

Core Viewpoint - Yann LeCun, a prominent AI scientist at Meta, is leaving the company to start a new venture focused on advanced machine intelligence, diverging from Meta's current investment in large language models (LLMs) [1][36][38]. Group 1: Departure and New Venture - Yann LeCun announced his departure from Meta after 12 years, stating that the company will be a partner in his new startup, although Meta will not be an investor [1][36]. - LeCun's new company will focus on teaching AI to understand the physical world rather than developing LLMs like ChatGPT [3][36]. Group 2: Critique of Large Language Models - LeCun has been a vocal critic of LLMs, arguing that they have reached their limits and lack true understanding of the physical world, memory, and multi-step reasoning capabilities [6][8]. - He believes that LLMs are merely token generators and do not possess the reasoning abilities necessary for true intelligence [6][20]. Group 3: The Concept of World Models - LeCun advocates for the development of "world models," which he believes are essential for achieving true machine intelligence, as they allow for understanding and interaction with the physical world [12][22]. - He emphasizes that human-like intelligence requires more than just language processing; it necessitates the ability to interact with and learn from the environment [35][36]. Group 4: Industry Implications - The AI industry is heavily focused on LLMs, which LeCun describes as a "black hole" that absorbs resources and attention, hindering progress in other areas of AI research [8][40]. - LeCun's departure and criticism of LLMs may signal a shift in the AI landscape, as he suggests that the next major breakthroughs will come from alternative approaches like world models [12][40].

Group 1 - The overall impact of AI on employment is characterized by four intertwined effects: enhancement, substitution, supplementation, and creation [3][4] - AI enhancement leads to widespread efficiency improvements, with a potential 15% increase in labor productivity in developed markets, while 25% of global jobs face risks from GenAI, with high-income countries seeing a 34% risk [3][4] - The substitution effect of AI is currently faster than the creation of new jobs, but this does not equate to mass unemployment, as companies are adopting strategies like hiring freezes and role transitions instead of large-scale layoffs [5][6] Group 2 - AI is expected to supplement labor in high-demand, high-risk jobs, addressing structural labor shortages, particularly in sectors facing challenges from an aging population [5][6] - The creation of entirely new job types is lagging, with existing roles increasingly requiring AI skills; positions demanding AI tool proficiency have grown by 68% year-on-year [6][20] - New job categories in the AI ecosystem can be classified into five core types: Enablers, Collaborators, Governors, Promoters, and Supporters, reflecting different value creation roles within the AI landscape [8][10][15] Group 3 - The emergence of new job characteristics includes deep specialization, cross-disciplinary integration, human-machine collaboration, and dynamic evolution of roles, indicating a shift in job nature and requirements [20][22][23] - AI-native jobs are expected to emerge primarily from technology companies, with a significant increase in AI-related job postings projected for 2025 [25] - The service industry is anticipated to be the main area for employment growth, driven by AI's integration into service roles and the increasing demand for jobs in elder care and community services [26][27] Group 4 - The shift towards flexible employment models is accelerated by AI, with a rise in gig work and one-person enterprises, as traditional job structures evolve into task-based systems [27][29] - Companies are encouraged to adopt people-centric AI transformation strategies, ensuring employee rights and providing retraining opportunities to adapt to AI integration [30] - A collaborative approach among government, enterprises, and workers is essential to create an employment-friendly environment, including support for AI innovation and adjustments to social security systems [31][32]