红杉中国合伙人周逵在开场致辞中，从AI技术进化、AI产品特征、AI公司特征、AI商业模式以及未来智能 公司的竞争态势和结果等多个维度，分享了他对AI当下发展与未来走向的思考和见解。他表示，AI是人类 技术进步的新里程碑， "具身"的含义好似给现实生活的各类存在都能带上"大脑"的机会。他说："无论 是'硬'的机器人还是软的'Agent'，共同特点都是在获得信息同时有进一步交付的能力。企业选择Leval 2还是 Leval 4的智能目标，导致的智能能力和商业结果大不相同。"他尤其期待看到"世界模型"的重要进展，期待 下一个AI智能的Aha Moment出现。 Genesis创始人及CEO周衔和红杉中国合伙人公元进行了连线对话。周衔表示，具身人工智能技术的发展， 大概率不会出现陡然的转折点。人们或许会目睹机器人逐步渗透进一些To B的应用场景，在这一阶段，它 暂时无需与人类开展复杂的交互。随着技术的经年打磨与渐次升级，其能力将得到稳步提升，逐步迈向家 庭领域，成为人们日常生活中的得力助手。若持乐观态度，机器人技术有望在约3年左右实现关键性突破， 迎来真正意义上的商业化转折。 红杉中国合伙人郑庆生在演讲中表示，目前， ...

Core Viewpoint - The article emphasizes "unconventional" entrepreneurial thinking as a means to identify and seize opportunities for success, suggesting that mindset plays a crucial role in responding to impending changes [2]. Group 1: "We Can Do It" - Companies should focus on their core competencies and be open to exploring new opportunities beyond their traditional capabilities, as demonstrated by Brazilian entrepreneur Arnold Correia, who successfully ventured into satellite broadcasting despite lacking prior knowledge [3][4]. Group 2: "Problem-Oriented, Not Product-Oriented" - Entrepreneurs should prioritize solving problems rather than merely focusing on product innovation. Jonathan Thorne's development of a non-stick surgical tweezer illustrates the importance of addressing significant issues in specific medical fields rather than just enhancing existing products [5][6]. Group 3: "Focus on Specific Markets" - Identifying and addressing needs within niche markets can lead to substantial opportunities. Phil Knight and Bill Bowerman's creation of Nike stemmed from recognizing the unique requirements of long-distance runners, leading to the development of specialized running shoes [7]. Group 4: "Boldly Ask for Money" - For startups, securing cash flow is vital. Elon Musk's approach at Tesla, where he sold 100 Roadsters before production, exemplifies the effectiveness of pre-selling to generate capital for development [8][9]. Group 5: "Borrowing Resources" - Entrepreneurs can leverage existing resources to minimize initial investment. Tristram and Rebecca Mayhew's Go Ape company successfully partnered with the Forestry Commission to utilize their parks for adventure courses, demonstrating the potential of collaborative resource sharing [10].

Core Viewpoint - The article discusses the launch of a new AI benchmark testing tool called xbench by Sequoia China, aimed at creating a more scientific and effective evaluation system for AI capabilities, particularly in real-world applications [1][2]. Group 1: xbench Overview - xbench employs a dual-track evaluation system that constructs a multidimensional dataset to track both the theoretical limits of AI models and the practical value of AI agents in real-world scenarios [2][3]. - The tool features an Evergreen Evaluation mechanism, ensuring continuous updates to testing content to maintain relevance and timeliness [2][3]. Group 2: Evaluation Methodology - The initial release includes two core assessment sets: xbench-ScienceQA for scientific question answering and xbench-DeepSearch for deep search capabilities, with comprehensive rankings of major products in these fields [3][19]. - The evaluation methodology focuses on aligning assessments with real-world applications, particularly in recruitment and marketing sectors, to establish clear business value [3][12]. Group 3: Historical Context and Development - xbench has been used internally by Sequoia China for over two years to track and evaluate foundational model capabilities, with significant improvements observed in model performance over time [5][7]. - The tool's question bank has undergone multiple updates to reflect increasing complexity and relevance to real-world tasks, demonstrating rapid advancements in AI model capabilities [5][7]. Group 4: Future Directions - The article emphasizes the need for innovative task settings and evaluation methods that align with practical applications, moving beyond traditional assessment frameworks [8][22]. - Future evaluations will focus on dynamic, real-world tasks that reflect the evolving needs of various professional fields, with an emphasis on collaboration with industry experts to refine assessment criteria [24][27]. Group 5: Long-term Evaluation Strategy - The Evergreen Evaluation approach aims to mitigate issues of question leakage and overfitting by maintaining a dynamic and continuously updated assessment pool [11][30]. - The article outlines a vision for ongoing assessments that adapt to the rapid evolution of AI technologies and their applications in diverse professional contexts [30][35].

而且，动态设计中也涉及大量叙事元素。 当视觉叙事与图形元素、策略性文案及艺术指导完美融合时，能 达到最佳的动态效果，从而打动受众。 让我们从一些案例开始看起，了解 动态设计可以放在哪些触点： 如今，视频贴片广告、社交媒体流广告、APP开屏广告、网页用户界面、微互动设计、数字广告牌、户外屏幕 ——一切都在动！ 屏幕无处不在，在线内容也呈井喷式增长。品牌纷纷调整策略，在坚守传统渠道的同时，积极拥抱数字化触 点。从沉浸式的品牌体验到流媒体和广播内容，动态设计已然成为很多行业的标配。在一个"大屏幕看着，小屏 幕刷着"已是常态的世界里，让内容不动才是奇怪的事情。 现在，让我们聚焦于品牌的"动态设计" （Motion Design，有时亦称"动效设计"） ，聊聊如何创造出能真正打动受 众的动态效果。 动态设计是一种通过动画技术让设计 （甚至更广义的"品牌"） "生动"起来的艺术与工艺。 "动态设计"的目的是持续且有效地传递信息，强化项目的概念和视觉语言——通常是为了鼓励用户采取行 动。其对象包括品牌标志 、 符号、字体、图片、视频片段、插画、图标——任何传递信息的视觉元素都可 以成为设计的切入点。 需要考虑的动态设计触点 ...

今天，我们再一次迎来国际生物多样性日。今年的主题是 "万物共生 和美永续" ，呼吁以人与自然和谐共生之 道，创和美永续之路，进而推进全球生物多样性保护治理新进程。 最早拥有三色视觉的动物是一类 非脊椎动物——节肢动物 （如昆虫、蜘蛛和甲壳类动物） 。而在4.2亿至5亿 年前， 脊椎动物 也开始具备三色视觉，使它们能够比单色视觉的生物更精准地识别猎物和捕食者，提高在 环境中的导航能力。 化石证据提供了进一步的线索。例如， 三叶虫 ——一类生活在5亿多年前的已灭绝的海洋节肢动物，就拥 有 复眼 。这一视觉结构能够探测多种光波长，使三叶虫在昏暗的海洋环境中获得进化优势，提高运动感知 能力和生存竞争力。 这些证据表明，生物在自身变得五彩斑斓之前，便已具备感知色彩的能力。 地球生命共同体的每一次脉动，都与人类命运息息相关。从灰棕、柔和的主体色彩到今日的五彩斑斓，从雄孔 雀炫目的羽毛到绚烂多姿的花朵，地球历经了一系列的进化——生命的色彩是基因密码的具象表达，是物种生 存的底层逻辑的体现。在这个过程中，生物多样性构筑了人类生存的根基。 但一个不可否认的事实是：全球众多物种的遗传多样性正在加速丧失，那些承载着适应潜能的基因宝 ...

Core Viewpoint - The article emphasizes the importance of human resource management from an organizational perspective rather than just a human resources function, highlighting that selecting the right people is paramount for long-term success [2]. Group 1: Importance of Employee Selection - Research indicates that the performance difference between top employees and average employees is significantly larger than expected, with top performers in various industries outperforming their peers by substantial margins [3]. - Choosing the wrong employees can lead to inefficiencies in subsequent HR functions like training and retention, making it crucial to select the right individuals from the outset [3]. Group 2: Investment Focus - Companies should shift their investment focus from training to recruitment, as the costs associated with hiring the wrong person can be much higher than those incurred during training [4]. Group 3: Five Rules for Improving Selection Probability - **Rule 1: Define Key Selection Criteria** Companies should clearly outline the essential characteristics they seek in candidates, as exemplified by Amazon's focus on practicality, ownership, and resilience [6]. - **Rule 2: Conduct Proper Assessments** Effective selection tools include cognitive ability tests, work sample tests, and structured situational interviews, while relying solely on personality assessments may not yield reliable results [8]. - **Rule 3: Implement Scientific Interviews** Structured situational interviews, which follow the STAR principle, are shown to be the most effective method for assessing candidates [10]. - **Rule 4: Emphasize Probation Periods and Background Checks** Utilizing probation periods allows companies to evaluate candidates in real work situations, and thorough background checks are essential, especially for higher-level positions [12]. - **Rule 5: Have the Courage to Replace Unsuitable Employees** Companies should be prepared to part ways with employees who do not meet expectations, as maintaining a harmonious relationship relies on mutual suitability [14][16].

有的时候你可能会有这样的迷惑：怎么新闻里的AI又写科研论文，又能自动驾驶，但是我们手里的AI工具总是 时灵时不灵呢？或许这份斯坦福大学最新的《2025年人工智能指数报告》能够解答你的困惑。 这份400多页的年度报告，包括了对不断演变的AI硬件格局的深入分析、对推理成本的全新估算，以及对AI领域 学术发表和专利申请趋势的新分析。同时还引入了有关企业采用尽责AI实践的新数据。我们编译了报告的部分 重要成果，希望能够帮助大家更好地理解AI技术的发展，充分利用它获得先发优势。 【温馨提示】文末"阅读原文"可下载原报告 人工智能日益融入日常生活 从医疗保健到交通运输的诸多领域，人工智能正迅速从实验室走向日常生活。2023年，美国食品药品监督 管理局 （FDA） 批准了223种配备人工智能的医疗器械，而2015年这一数字仅为6种。在道路上，自动驾驶 汽车已不再处于试验阶段，某自动驾驶汽车运营公司已每周提供超过15万次自动驾驶出行服务。 在这波人工智能大热潮中，科学和医学领域涌现了令人惊喜的新气象。 多个新发布的基础模型，它们将助 力材料科学、天气预报和量子计算等方面的研究。 许多公司正尝试将AI的预测与生成能力转化为有利 ...

Core Viewpoint - The article discusses the rapid advancement of AI-generated content across various forms such as text, images, and videos, emphasizing the need for individuals to develop skills to discern between human-created and AI-generated content [5][24]. Group 1: Identifying AI-Generated Text - AI-generated text often exhibits a distinct "flavor," characterized by overly precise language and emotional dilution, making it easier to identify [8][10]. - Common traits of AI writing include excessive use of complex vocabulary, a barrage of examples and metaphors, and a lack of personal experience or unique insights [9][10]. - AI text tends to be overly polished and consistent, lacking the natural rhythm and emotional fluctuations typical of human writing [9][10]. Group 2: Identifying AI-Generated Images - AI-generated images can be scrutinized for key details such as hands, teeth, and eyes, which are common areas where AI makes mistakes [12][13]. - Consistency and logic in lighting, shadows, and background elements are crucial for identifying AI images; discrepancies can indicate AI generation [15][17]. - Observing texture and symmetry can also reveal AI-generated images, as they may appear unnaturally smooth or overly perfect [17]. Group 3: Identifying AI-Generated Videos - AI-generated videos often struggle with replicating human facial expressions and may exhibit unnatural eye movements or facial symmetry [19][20]. - Illogical actions in videos, such as the absence of typical human habits, can signal AI involvement [20][21]. - Trusting one's intuition about the overall feel of a video can be a valuable tool in identifying AI-generated content [21]. Group 4: Tools for Detection - Various AI detection tools are available to analyze text, images, and videos for signs of AI generation, including Grammarly, ZeroGPT, and deepfakedetector.ai [23][24]. - No single detection tool is 100% accurate; combining multiple methods and tools is recommended for better reliability [24]. - The ongoing evolution of AI technology presents a continuous challenge in distinguishing between human and AI-generated content, necessitating critical thinking and media literacy [24].

提升人类科研效率，该AI上场了 五一期间，FutureHouse推出四个AI科研智能体，官方用四种动物来命名它们，分别是Crow （通用智能体） 、Falcon （自动化文献综述智能体） 、Owl （调研智能体） 、Phoenix （实验智能体） 。这些AI智能体可访问 完整科学文献全文，还具备信息质量评估能力。 图源：FutureHouse官网 这其中，Crow、Falcon和Owl通过了严格的基准测试，在搜索精度和准确性上已经超越了o3-mini、GPT- 4.5、Claude-3.7等当下顶级搜索模型。这三个模型 可以访问大量完整的科学文本 ，这就意味着，你可以向 它们提出关于实验方案和研究局限性的更详细的问题。它们还能使用各种因素来区分来源质量，确保它们 不会依赖低质量的论文，或者是流行的科学来源。 目前这四个科学家智能体虽然还不能自主完成大多数科学研究，但是人类已经可以用它来生成和评估新的 假设，规划新的实验——速度要比以前快很多。 而且这些智能体的推理过程完全透明，对每个信息来源都 进行了多阶段的深入分析。 更重要的是，用户可以清晰地查看整个推理过程，了解智能体得出结论的每一 步依据。 再加上Fu ...

Core Viewpoint - The article emphasizes the importance of personal branding (personal IP) for entrepreneurs and company leaders, highlighting how it can enhance trust and connection with customers, suppliers, and investors [2][5][6]. Group 1: Importance of Personal IP - Entrepreneurs and founders serve as the most persuasive "human billboards" for their brands, influencing consumer perceptions through their personal stories and professional expertise [2][5]. - A strong personal IP can humanize a company, making it more relatable and trustworthy to consumers, especially in a market where traditional advertising is losing effectiveness [5][6]. - Research indicates that 63% of respondents feel companies with active social media presence from their CEOs appear more "warm," and 64% appreciate personal content shared by CEOs [5][6]. Group 2: Building Personal IP - Defining personal IP involves understanding one's true self and sharing personal experiences, education, and interests to resonate with potential customers [8][9]. - Targeting the right social media platforms is crucial; focusing on one or two platforms can yield better results than spreading efforts too thin [10]. - Continuously showcasing value through insightful content and personal stories can attract and retain audience attention [11][12]. Group 3: Networking and Consistency - Establishing connections with colleagues, clients, and partners is vital for building a personal IP, as recommendations and endorsements can enhance credibility [14]. - Consistent content updates and maintaining a unified brand image across all platforms are essential for reinforcing personal IP and professional credibility [15]. Group 4: Employee Personal IP - Employees' personal IPs can collectively shape a company's overall image, making it more approachable and relatable to potential clients [19]. - Engaging employees in personal branding efforts can significantly increase interaction rates on social media, often outperforming official company accounts [19].