Core Viewpoint - The rise of "bed-sharing rental" among young people in major cities is a response to high rental costs, allowing them to save money while compromising on personal space and privacy [1][2][4]. Group 1: Economic Impact - Bed-sharing rental allows individuals to split the cost of rent, reducing monthly expenses significantly; for example, a room that costs 3000 yuan can be shared, resulting in each person paying 1500 yuan [6][10]. - The demand for affordable housing options under 1500 yuan per month is increasing in first-tier cities, leading many young professionals to opt for bed-sharing arrangements [12][34]. Group 2: Social Dynamics - Young individuals often form friendships through bed-sharing, as seen in the experiences of Yangyang and her roommate, who became friends after initially meeting online [5][18]. - The arrangement can provide emotional support, helping alleviate feelings of loneliness in a new city [18][19]. Group 3: Privacy and Safety Concerns - Sharing a bed with a stranger raises significant concerns regarding personal safety and privacy, as individuals must navigate shared living spaces carefully [3][20]. - There are documented cases of theft and fraud in bed-sharing situations, highlighting the risks involved [20][21]. Group 4: Lifestyle Adjustments - Individuals in bed-sharing arrangements often have to adapt their daily routines to accommodate their roommates, such as being mindful of noise levels and personal space [26][22]. - The experience of bed-sharing can lead to a lack of personal freedom compared to living alone, as individuals may feel restricted in their actions [26][29]. Group 5: Long-term Implications - Many young professionals view bed-sharing as a temporary solution while they seek more stable housing options, reflecting the broader challenges of urban living [35][36]. - The high cost of living in major cities continues to push young people into unconventional living arrangements, impacting their overall quality of life and future housing decisions [33][34].

Core Viewpoint - The article emphasizes the importance of independent thinking in an AI-driven world, arguing that true independent thought goes beyond merely consuming information and requires deeper emotional and cognitive engagement [1][2]. Group 1: Understanding Independent Thinking - Independent thinking is not merely about information acquisition or following trends; it involves pausing to critically analyze and reflect on information [3][4]. - Genuine independent thought requires self-inquiry before expressing opinions, asking oneself if they truly understand the subject matter and what their personal feelings are [5]. Group 2: The Role of Emotions in Thinking - The foundation of independent thinking is emotional awareness, where initial emotional reactions to information serve as a starting point for deeper inquiry [6][7][8]. Group 3: Critical Thinking - Critical thinking involves recognizing the boundaries of arguments rather than simply criticizing them; it requires a thorough examination of viewpoints [9]. Group 4: AI as a Tool - AI should be viewed as a tool to enhance thinking rather than a replacement for it; individuals should engage in self-reflection before consulting AI [10][12]. Group 5: Developing Inner Awareness - There is a need to cultivate an "inner sense" to better understand one's emotions and thoughts, which can enhance independent thinking [13]. Group 6: Exercises to Enhance Thinking - The article suggests four exercises to foster active thinking: keeping an emotion diary, challenging accepted viewpoints, creating a question list before using AI, and practicing mindfulness meditation [14][17]. Group 7: Conclusion - Despite AI's capabilities, it cannot replicate personal experiences, emotions, and insights, which are essential for genuine thought processes [18][19].

很多人说，"2025 年是 AI agent 元年"。各种新闻文章标题都这么写： 而我却刚刚花了一年时间搞清楚哪些 Agent 在生产环境里真正能用，也正因此，我才不看好这股风潮。 一、我不是唱反调的人，我是真干过的 过去一年，我做了十几个上线的 Agent 系统，覆盖整个软件开发流程，比如： 这些系统确实能用，也确实创造了实际价值，每天都能帮人省下好几个小时的手动操作。也正因如此，我才认为，外界把 2025 年称作 "AI Agent 元年" 的 说法，忽略了很多关键现实。 二、要点速览：关于 AI Agent 的三个残酷现实 在构建了 12 套以上的生产级系统之后，我得出了以下几点结论： 三、一个没人愿意面对的数学现实 所有做 AI Agent 的公司都在回避一个难以接受的事实：在生产级别的多步骤任务中，错误的累积让"全自动智能体"在数学上根本行不通。 "AI agent 会彻底改变工作方式"； "Agent 是 AI 的下一个风口"； "未来属于 Agent"。 开发类 Agent：自然语言生成 React 组件、重构老代码、自动维护 API 文档、根据说明生成函数。 数据和基础设施类 Agent：自动 ...

上周四的线下活动圆满结束，大家很热情，嘉宾很卷，干货很满。有AI、有宏观，还有现场的AI agent demo。整体很精彩，有一种经济上行期的美。 直到尾声Q&A环节，问题又把我们拉回到了经济下行期的现实。 大家主要关心的是： 1. 百年未有之大变局再叠加势不可当的AI，年轻人到底还有什么机会？ 2. 那怎么办？ 一 线性回归这道题解不出来了，怎么办？ 每一代人都有自己的挑战。对00后（Z世代）来说，挑战叫做：所有和人生规划相关的线性回归不work了。比如： 卷教育线性回归不出"就业"，或者回归的结果是"灵活就业"。 工资收入线性回归不出"买房"和退休。 其实美国的Z世代也一样，目前美国人买房的中位年龄大约是54岁，在1980年代是34岁。藤校毕业也可能找不到工作。 来感受下美国毕业生的失业率： 不只是中国和美国，欧盟、英国、加拿大、日本都一样： 最直接的情绪是迷茫。不用我说，现在大家都应该感受到了，各种文章讨论年轻人的精神状态，虚无主义蔓延。其实都不用看文章，路上随便擦肩而过一 个Z世代，迷茫和困惑就扑面而来——真不知道该怎么办。 于是他们被冠名为"被拒一代"（the most rejected gener ...

Group 1 - The conflict between Thailand and Cambodia escalated with military engagement near the Ta Moan temple, leading to both sides deploying heavy artillery and airstrikes [1][3][10] - Thailand's military response included airstrikes using F-16 fighter jets, targeting Cambodian military facilities, which resulted in significant damage [3][22] - Cambodia retaliated with artillery fire, including the use of BM-21 rocket launchers, causing civilian casualties in Thailand [5][10] Group 2 - The historical context of the conflict centers around the disputed ownership of the Preah Vihear temple, with both countries claiming sovereignty based on historical treaties and court rulings [12][14][16] - The region is rich in strategic resources, including rare earth minerals and oil fields, which adds to the tension between the two nations [19][20] - Domestic political pressures in both countries are influencing the conflict, with leaders seeking to bolster their support through nationalistic rhetoric [20] Group 3 - The military tactics employed by both sides reflect modern warfare trends, with Thailand utilizing drones for reconnaissance and targeting, similar to strategies seen in the Russia-Ukraine conflict [22][24] - Cambodia's approach relies on traditional artillery tactics, but their lack of advanced military hardware limits their effectiveness against Thailand's superior air capabilities [30][31] - The ongoing conflict highlights the evolving nature of warfare in the region, with both sides adapting to new technologies and strategies [21][32]

流量是做自媒体的基础变量。没有数据反馈你无法优化内容，也无法验证定位。 但长期看价值，是因为你给观众带来的价值感知，才能发展出信任关系和商业转化。 一个成熟的内容创作者，不逃避流量，也不迷信爆款，而是需要清楚地知道不同阶段自己在获取什么。 二 平台的核心推荐机制是基于短期反馈的：播放量、完播率、点赞评论分享的数据模型决定你的分发。 本文来自微信公众号：珞石居，作者：珞石居，原文标题：《个人的影响力是均值回归的》，题图来 自：AI生成 短期看流量，长期看价值。 三 要用短期流量验证内容结构，再用长期内容塑造价值认知和沉淀影响力。 一 内容行业的长期回报，不是来自流量的总量，而是来自流量的质量和转化。 很多账号做生活段子、情绪共鸣类内容，数据非常高，但商业化（广告、带货、咨询、课程、付费社 群）的转化很差。因为观众没有认可你在某个领域有高价值，自然也不信任你。 所以短期你要靠爆款抢到注意力；但长期来看，你必须通过内容去建立明确的价值锚点。 这个机制当然主要推送高情绪刺激、信息密度高、结构紧凑的内容。 所以一些账号会突然暴涨几百万播放，但几天后热度迅速消失，新内容也很可能没人看。 原因很简单：平台是推内容的，不是推 ...

当辛顿在上海人工智能大会上，把AI比作人类正在豢养的一只可爱的小老虎时，我想起了机器人先驱莫拉维克 的名著《心灵的孩子：机器人与人类智能的未来》（Mind Children:The Future of Robot and Human Intelligence），他把人工智能看成是人类教养自己的智能后代，这甚至可能是一种人类文明的继承者。 在未来半个世纪里，计算能力将持续增长，机器人也将逐步超越"爬行动物阶段"：它们将像哺乳动 物那样学习、像灵长类动物那样构建对世界的模型，最终达到类人推理的能力。 那时，取决于你的视角，人类要么造出了一个真正值得的继任者，要么超越了自身的某些先天局 限，从而将自己转化为一种全新的存在。 莫拉维克并非无视AI可能带来的风险，因为它会代替人类，但是莫拉维克更倾向于认为这是对人类的一种"温和 的接管"，这是人类演化和自然选择的产物。 这也体现出莫拉维克与辛顿对于AI发展各自的侧重。莫拉维克推崇进化时间带来的感知/动作优化；强调机器模 拟人脑需大量计算力与架构复制。而辛顿注重训练方法与神经结构设计，强调数据与算法的力量；更关注如何 让AI学习而非进化。一个强调进化的时间，一个强调设计的 ...

7月26日，世界人工智能大会（WAIC）在上海正式拉开帷幕。 在展会现场，具身智能无疑是最备受瞩目的行业。从灵活完成格斗动作的人形机器人，到能精准执行家务、康复辅助的服务型机器人，这些具备物理交互能 力的智能体凭借流畅的动作、自然的交互赚足了观众眼球。 然而，在这场科技盛宴中，当人们沉浸在具身智能带来的震撼体验时，人工智能产业的核心底座——以芯片、板卡、服务器、计算集群等为核心构成的算力 基础设施，依然是整个行业绕不开的关键话题。 与往届大会相比，今年算力基础设施展区呈现出截然不同的风貌。 往年展台上随处可见的"参数竞赛"已悄然淡去，各厂商送展的芯片和服务器展品中，很少再有将某项极致性能参数用醒目字体刻意标记出来的情况。 曾经被重点标注的技术指标，如今更多地融入到具体的产业解决方案中，或是隐藏在真实的应用场景演示里。 现场的讨论也多是基于"碎片化算力资源统筹"、"低功耗与低成本"与"垂类产品软硬件整合"这些非常务实的话题。 全链路国产化进行时 在算力基础设施的务实转型浪潮中，"全链路国产化"的推进节奏尤为引人关注。 过去几年，全球供应链波动引发的芯片断供风险，以及核心技术"卡脖子"的现实挑战，反复挑动着国内人 ...

Core Insights - The global advertising industry is undergoing a significant restructuring by 2025, with two distinct paths emerging: one focusing on scale through mergers and acquisitions, and the other leveraging technology and AI for efficiency and agility [2][3][4] - The competition between scale and speed represents a strategic choice that can determine the success or failure of companies in the industry [4] Company Performance - Public financial reports reveal contrasting performances between major players: Publicis Groupe (阳狮) reported a net income of €3.617 billion in Q2, a year-on-year increase of 4.6%, while WPP experienced a revenue decline of 5.5% in the same period [7][8] - Publicis has raised its organic revenue growth forecast to nearly 5%, while WPP has downgraded its annual forecast to a negative growth of 3% to 5% [7][8] Strategic Choices - Publicis has made significant investments in technology and data, positioning itself as a platform-driven growth company, while WPP has struggled with debt reduction and brand integration, appearing to be reactive rather than proactive [15][31] - The acquisition of Sapient in 2014 marked a pivotal moment for Publicis, allowing it to integrate technology and digital transformation into its core services, while WPP's strategy remained conservative, focusing on optimizing traditional research businesses [17][30] Data and Technology - The acquisition of Epsilon for $4.4 billion allowed Publicis to establish a stronghold in first-party data, which is becoming increasingly critical in the advertising landscape due to privacy regulations and the decline of third-party cookies [32][35] - Epsilon's data capabilities have significantly enhanced Publicis's revenue structure, with digital income rising from less than 30% to over 50% [47] AI and Platformization - Publicis has introduced the Marcel platform to enhance internal collaboration through AI, while also investing €300 million over three years in AI research and data platform upgrades [56][63] - WPP has also developed its own technology platform, OpenX, but lacks the proprietary data assets that Publicis has secured, limiting its competitive edge [65][51] E-commerce and Retail Media - Publicis has recognized the shift towards retail media advertising, acquiring companies like Profitero and Corra to enhance its capabilities in e-commerce solutions [76][77] - Retail media advertising is projected to account for over 20% of global ad spending by 2024, indicating a significant growth opportunity for companies that can adapt quickly [74] Industry Dynamics - The advertising industry is facing competition not only from traditional players but also from consulting firms and tech giants, which are increasingly encroaching on advertising budgets [84][85] - The success of Publicis raises critical questions about whether advertising holding companies must evolve into technology companies to remain relevant in the changing landscape [87]

林远是TOP3高校的学生。在这个暑假，她回到自己老家所在的南方省份地级市，做本校的招生志愿者。 招生组的目的是在普通批分数段内，争取高分生；在低分数线的冷门专业内，招到足够数量的学生。往常，后者并非难题。招生组只需要用惯常的策略 ——描摹蓝图，勾画理想，强调名校光环——就能完成任务。 但林远发现，今年，招生组的策略正逐一失效。越来越多的考生开始意识到，志愿填报在个人意愿之外，是多方力量的博弈。 在高校招生组的任务、家庭的期许、中学的绩效之间，刚成年不久的孩子们在多方利益中周旋。有人识破被营造的幻觉，尝试从各类期许中逃逸。但离开 主流描摹的道路后，地域、阶级与信息差筑造的鸿沟，又给他们的前路带来新的迷雾。 以下是林远的自述： 一、策略失效 "你好，我是来问我的分数能不能报你们学校的。"房间里百无聊赖的氛围被座机的铃声搅碎。电话那头，女孩的声音怯生生的。 在记录下她的成绩和毕业学校之前，我就大概知道了不能给到肯定的回复：对于我所在的这所TOP3高校招生组，主动打来电话的考生，往往都游离在分 数线上下，并不是有竞争力的高分考生。 我建议对方参考以基础学科为主的"强基计划"和提前批中的小语种专业。这两个方向的分数线更 ...