在人工智能的助力下，全球实现或者正在从"白手起家"跨越到亿万富豪行列的人越来越多。 2025 年 4 月 23 日，《福布斯》首次发布 "全球最年轻的白手起家女性亿万富豪" 排名，人工智能公司 Scale AI 的联合创始人 Lucy Guo（郭露西）以 30 岁 年龄、约 12.5 亿美元的净资产，力压美国歌手Taylor Swift（泰勒・斯威夫特），位居榜首。 | 姓名 | 走龄 | 身家 | 国籍 | 财富来源 | | --- | --- | --- | --- | --- | | Lucy Guo | 30 | 12.5 | 美国 | 人工智能 | | 泰勒·斯威夫特(Taylor Swift) | 35 | 16 | 美国 | 音乐 | | 丹妮拉·阿莫迪(Daniela Amodei) | 37 | 12 | 美国 | 人工智能 | | 梅兰妮·珀金斯(Melanie Perkins) | 37 | 57 | 澳大利亚 | 软件 | | 蕾哈娜(Rihanna) | 37 | 14 | 巴巴多斯 | 化妆品、音乐 | | 卢依雯 | 37 | 11 | 中国 | 珠宝 | | | (单位:亿美元 ...

该工作由通用人工智能研究院 × 北京大学联手打造。第一作者郑欣悦为通用人工智能研究院研究员，共同一作为北京大学人工智能研究院博士生林昊苇， 通讯作者为北京大学助理教授梁一韬和通用人工智能研究院研究员郑子隆。 开发能在开放世界中完成多样任务的通用智能体，是 AI 领域的核心挑战。开放世界强调环境的动态性及任务的非预设性，智能体必须具备真正的泛化能力 才能稳健应对。然而，现有评测体系多受限于任务多样化不足、任务数量有限以及环境单一等因素，难以准确衡量智能体是否真正 「 理解 」 任务，或仅是 「 记住 」 了特定解法。 为此，我们构建了 Minecraft Universe （ MCU ） —— 一个面向通用智能体评测的生成式开放世界平台。 MCU 支持自动生成无限多样的任务配置，覆 盖丰富生态系统、复杂任务目标、天气变化等多种环境变量，旨在全面评估智能体的真实能力与泛化水平。该平台基于高效且功能全面的开发工具 MineStudio 构建，支持灵活定制环境设定，大规模数据集处理，并内置 VPTs 、 STEVE-1 等主流 Minecraft 智能体模型，显著简化评测流程，助力智 能体的快速迭代与发展。 开放世界 ...

Group 1 - AGI (Artificial General Intelligence) is defined as an AI system capable of matching human abilities across a wide range of cognitive tasks, characterized by its versatility and performance [4][11][12] - The development of AGI is seen as a continuation of the trend towards agent-based AI, where AGI serves as the "brain" of multiple agents, rather than being a standalone system [5][6] - The potential risks associated with AGI include existential threats to humanity, particularly if AGI systems are allowed to interact with the environment without strict limitations [12][13][14] Group 2 - The article emphasizes the importance of limiting AGI agents to narrow environments, ideally within a single team or organization, to minimize negative impacts [22][23] - It suggests that AGI should not operate globally, as this could lead to uncontrolled access to vast amounts of data and tools, increasing the risk of unintended consequences [19][20] - The design of AGI agents should prioritize local deployment within teams, allowing for collective training and supervision, which enhances safety and effectiveness [48][49] Group 3 - The article discusses the potential for AGI agents to be integrated into workplace environments, enhancing collaboration and efficiency while maintaining human oversight [28][30] - It highlights the advantages of multi-agent systems, which can solve complex problems through collaboration, specialization, and modularity, making them more adaptable and cost-effective [40][41][42] - The deployment of AGI agents should focus on team-level applications rather than individual use, to ensure that human decision-making and critical thinking skills are preserved [27][32]

每经记者｜张蕊 每经编辑｜陈旭 没有安全，数据流通就无从谈起；没有流通，数据就难以赋能千行百业。 5月10日，以"数智无界 安全共生"为主题的2025中国数谷·西湖论剑大会在杭州开幕。 中国计算机学会（CCF）原理事长、CCF计算机博物馆馆长梅宏在大会主论坛上提到，没有数据就不可能有智能，就像燃料和火箭的关系一样。现在业界经 常讲本轮AI革命的三要素：算法、数据、算力，实际上数据是关键。 推动AI（人工智能）向AGI（通用人工智能）演进，数据与AI的深度融合非常重要，但现在缺乏高质量的数据集致数据流通受限，这是不是一个关键的堵 点？ 对于《每日经济新闻》记者提出的这一问题，安恒信息技术股份有限公司（SH688023，股价49.18元，总市值50.30亿元，以下简称安恒信息）高级副总裁、 研究院院长王欣表示，数据的流通受限确实是一个关键问题。 私有数据是让模型落地到场景里发挥价值的关键因素 王欣对《每日经济新闻》记者表示，整个模型从构建到应用落地分为两个方面：一方面，现在"大厂"做一些基础的通用模型训练，更多用到的是互联网的网 页数据。这里存在一些具体数据层面的安全问题，主要是数据质量问题，这影响了模型本身的能 ...

Core Viewpoint - The AI industry is currently engaged in discussions about the adequacy of existing large model testing sets, with a consensus emerging that a new, universally accepted testing framework is needed to accurately assess the capabilities of advanced AI models [1][6]. Group 1: Current State of AI Testing - The article highlights that mainstream AI models have reportedly passed the Turing test, suggesting they meet the standards for Artificial General Intelligence (AGI) [1]. - Existing testing sets, such as MMLU, have been criticized for their inability to effectively evaluate the rapidly evolving capabilities of large models, leading to concerns about their reliability [3][4]. - The emergence of "cheating" practices, where developers manipulate testing sets to achieve higher scores, has further undermined the credibility of current evaluation methods [3][4]. Group 2: New Testing Initiatives - OpenAI has introduced the FrontierMath testing set, which shows significant performance differentiation among models, with the latest o3 model achieving a correct rate of 25%, far surpassing other models [5]. - However, concerns have been raised regarding OpenAI's access to the FrontierMath question database, which has led to questions about the integrity of this testing set [5]. - Industry stakeholders, including Scale AI and CAIS, are collaborating to design a new model testing set that aims to be more reliable and accepted across the board [6].

闫德利 腾讯研究院资深专家 一、AI已在诸多任务领域超越人类 AI发展日新月异，在许多任务上已经陆续超越人类基线水平。如2015年图像分类，2018年中等水平阅读 理解，2020年视觉推理、英语语言理解，2023年多任务语言理解、竞赛级数学，2024年博士级科学问 题。下图所示的8项关键任务技能中，AI仅在多模态理解和推理能力上还略逊人类一筹，但从2023年开 始就加速提升。我们有望很快见证AI 能力在现有主流基准上"全部超越人类水平"的奇点时刻。 图 选定的 AI 指数技术性能基准与人类表现对比 二、AGI的终极目标或于年内实现 我们已经构建了无数在特定任务上超越人类水平的AI系统，但它们缺乏通用性，无法应对超出预定任务 之外的问题，尚处于"狭义人工智能 （Narrow AI） "阶段。随着AI性能的大幅提升，具备跨领域能力、在 多个方面媲美甚至超越人类的、更强大的AI被提上日程。 人们常将之命名为"通用人工智能（AGI）" 。 各国高度重视AGI。2023年4月28日中共中央政治局会议提出："要重视通用人工智能发展"；英国《国家 人工智能战略》 （2021 ） 对AGI进行了专门强调，指出"必须认真对待A ...

Core Insights - The article discusses the evolution of AI models using games as a testing ground, highlighting the recent achievement of Google's AI model Gemini 2.5 Pro in independently completing the original Pokémon game, which has reignited interest in AI capabilities [4][30]. Group 1: AI Development and Gaming - AI has been tested through games for nearly a decade, with notable milestones including AlphaGo's victory over human players in Go and DeepMind's success in games like DOTA2 and StarCraft II [2][3]. - The use of games as a benchmark for AI intelligence remains prevalent, as demonstrated by Gemini's recent accomplishment, which was celebrated by Google's CEO and DeepMind's head [4][5]. Group 2: Challenges in AI Learning - The Moravec's paradox suggests that tasks perceived as easy for humans can be significantly more challenging for AI, which is exemplified by Gemini's achievement in Pokémon [6][7]. - The process of AI learning in games like Pokémon is complex, requiring the AI to develop its own understanding and strategies without predefined rules or guidance [16][17]. Group 3: Comparison of AI Models - Anthropic's Claude 3.7 struggled to progress in Pokémon, achieving only three badges after a year of iterations, while Gemini completed the game with approximately 106,000 actions, significantly fewer than Claude's 215,000 actions [11][30]. - The differences in performance between Claude and Gemini are attributed to their respective frameworks, with Gemini's agent harness providing better input processing and decision-making capabilities [34][35]. Group 4: Implications for AI Research - The ability of AI to navigate and complete games like Pokémon indicates its potential for independent learning and problem-solving in real-world scenarios [37][38]. - The choice of Pokémon as a training ground reflects the game's themes of growth, choice, and adventure, paralleling the journey of AI in understanding complex rules and environments [39][40].

近期 Founder Park 拜访了新创基金 Creekstone Ventures 的两位合伙人钟陆欢和李一豪。对于宏 观周期，技术周期，应用方向的洞见，和时代特性的创业者，开诚布公的交流。 理想，好奇，真诚，驱动着 Creekstone 在变化中探索，为更多优秀创始人服务和助力，也期待与 更多心怀梦想的同行人 Founder Park： 「减法」怎么理解？ " 01. 新基金聚焦 AI，和创始人连接更紧密 Founder Park： 介绍一下新基金目前的情况。 Creek Stone钟陆欢： 目前基金设立已基本完成，如果顺利，计划月底进行首次募集。这期基金规模预计 在数千万美金。 Founder Park： 现阶段有已经确定要投资的项目吗？ Creek Stone钟陆欢： 有的，目前有两个项目已经在协议阶段，分别是一家 AI Coding 公司和一家 AI 眼镜 公司。 我们之前在弘毅就投过 ToC 的 AI Coding项目，在企业级上也一直很看好。这次投的是一个ToB的企业级 Coding，我们和创始人关系不错，他刚从大厂离职出来，时机也合适，就决定作为新基金的第一个投资 项目。几家机构一起投的。至 ...

Core Insights - OpenAI is revising a multi-billion dollar partnership agreement with its largest investor, Microsoft, to ensure the potential for future IPO while granting Microsoft continued access to AI technology [1][2] - The negotiations focus on how Microsoft can secure equity in the restructured company that matches its $13 billion investment, amidst OpenAI's transition from a non-profit to a commercial entity [1][2] - OpenAI's CEO, Sam Altman, aims to develop artificial general intelligence (AGI) that surpasses human capabilities, while the company is also considering a shift to a public-benefit corporation model [2][3] Group 1 - Microsoft is cautious about OpenAI's restructuring plans, which aim to detach from its original non-profit mission and focus on commercialization [1][2] - The current agreement, effective until 2030, includes Microsoft's rights to OpenAI's intellectual property and revenue sharing [1] - Microsoft has proposed to relinquish some equity in OpenAI's new profit-driven ventures in exchange for rights to new technologies developed after 2030 [1][2] Group 2 - OpenAI's shift towards a public-benefit corporation is seen as a response to investor demands and is intended to facilitate future IPO opportunities [2][3] - Tensions have arisen between OpenAI and Microsoft due to differing operational styles, with OpenAI seeking autonomy in its business decisions [2] - Despite challenges, OpenAI executives believe that investor support will continue even if restructuring is delayed [3] Group 3 - OpenAI must demonstrate to California and Delaware authorities that its restructuring aligns with its public mission, with Delaware's Attorney General indicating a review of the new plan [4] - Legal challenges from former investor Elon Musk threaten to complicate the restructuring process, as he accuses the company of misappropriating charitable assets [3][4]

完整早报音频，请点击标题下方小耳机收听 【消息人士称马云回归是绝不可能的，马云是创始人也从来没离开过】 5月10日，记者从多位阿里巴巴内 部人士处获悉，阿里巴巴已全面打通了内网（内部论坛）权限，并调整员工跨业务流动机制、启动工牌 焕新。5月9日，针对"马云5月10日回归阿里巴巴，将要重启大集团模式"相关消息，记者向阿里巴巴集团 求证，截至发稿，暂无回应。阿里巴巴内部人士向记者表示，"马云回归是绝不可能的，并且马云是创始 人也从来没离开过"。（贝壳财经） 【苹果渠道官方调价：iPhone16 Pro最高降价176美元，Pro Max全容量降价160美元】 苹果5月10向渠道 商下发了调价通知，这也是苹果首次在周六宣布调价，给相当大一部分批发商打了个措手不及。苹果 iPhone16 Pro Max所有容量版本降价160美元，对应1313.06元人民币；而 iPhone16 Pro的128GB版本降价 176美元（对应1445.27元人民币），其他版本同样降价160美元。今年的618活动将于5月13日开始，因此 有渠道商认为苹果此次调价就是在为即将到来的618大促做准备。（IT之家） | | | SI Rebate ...