Core Insights - The development of AGI (Artificial General Intelligence) is a long-term process, currently transitioning from the "challenger" stage (L2) to the "reasoner" stage (L3), focusing on making AI actionable rather than just thinking [1] Group 1: Structural Changes in AI - The focus of AI competition is shifting from model capability to implementation capability, emphasizing the importance of models that can effectively deliver results [2] - The value positioning of AI is changing from tool efficiency to result closure, where AI applications should proactively complete tasks rather than just respond to commands [2] - The product form of AI is evolving from cloud computing to personal integration, requiring AI products to be embedded in user devices and daily life rather than remaining on servers [4] Group 2: Challenges in AI Implementation - There are three main challenges in AI implementation: 1. Scene integration, where AI struggles to operate efficiently across different devices, leading to fragmented user experiences [4] 2. Performance limitations, with current edge computing capabilities being insufficient to support models with over 7 billion parameters [4] 3. Trust issues, including privacy and ethical concerns, with a lack of standardized solutions for data rights and encryption [4] Group 3: Collaborative Initiatives - To accelerate AI implementation and achieve demand closure, a call for the establishment of an open AI terminal ecosystem alliance based on the GSMA platform has been made, aiming to unite AI model companies, operators, terminal enterprises, and internet companies [9] - The collaboration will focus on joint technological innovation, standard definition, business innovation, and talent cultivation to tackle AI implementation challenges [9] Group 4: Upcoming Product Launch - The company plans to launch the AI folding flagship Honor Magic V5 on July 2, which is expected to be the lightest folding screen phone globally and the "strongest" AI smart device in the industry [9]

Core Insights - The era of multimodal agent AI has just begun, with significant technical challenges remaining to achieve Artificial General Intelligence (AGI) [1] - Multimodal large models integrate various inputs and outputs such as text, speech, images, and videos, enhancing processing capabilities and user interaction experiences [3] Multimodal Technology Development - Multimodal technology is essential for achieving AGI as it provides richer contextual understanding and improves model performance and accuracy [3] - The technology can be categorized into understanding and generation tasks, with challenges in modality encoding alignment and high-quality content generation [3][4] Technical Evolution - The current multimodal understanding models are primarily based on pre-trained large model technology, with differences mainly in connector design and modality alignment methods [3] - Multimodal understanding models mainly focus on visual and language aspects, with aspirations to handle more modalities in the future [3] Future Directions - Future multimodal large models are expected to unify understanding and generation, although key technologies such as backbone network design and modality alignment still require further research [4] - The industry remains in its early stages, but there is confidence in the application prospects of multimodal technology in fields like search, creation, and robotics [4]

6月19日，荣耀CEO李健在2025世界移动通信大会（上海）上发表《开放共生，众木成林让AI走进生 活》主题演讲，详谈了AGI（通用人工智能）时代下AI发展的当下和未来。 在李健看来，AGI（通用人工智能）是一个长期演进的过程，当前处于从推理者到行动者发展的过渡阶 段，AI正在完成从"思考"到"行动"的关键跨越，当务之急是要"让AI能解决问题"，只有让AI落地生根， 让用户日常使用起来，才能真正释放AI的潜力，AI围绕"落地"正在发生三个结构性转变。 李健表示，他比较认可Sam Altman对AI五个阶段的划分。从L1到L5，从聊天者，到推理者，再到行动 者、创新者，最后到组织者。从"会说"，到"会想"，再到"会做"、"会创"，最后到"会管"。AI的发展， 分五个阶段逐级进化，最终实现AGI。当前，人类正处在L2与L3之间，AI正在从"思考"向"行动"跨越。 "AGI一定是一个长期演进的过程，当务之急是要'让AI能解决问题'，只有让AI落地生根，让用户日常使 用起来，才能真正释放AI的潜力。基于此，我们认为AI围绕'落地'正在发生三个结构性转变。"李健 说。 在李健看来，当前AI落地，仍存在三大挑战，场景、性 ...

Core Viewpoint - The CEO of Honor, Li Jian, emphasized the importance of making AI practical and accessible in daily life, advocating for the establishment of an AI terminal ecosystem alliance to accelerate AI integration into everyday use [1][3]. Group 1: AI Development and Strategy - Honor's CEO outlined the evolution path of AI, stating that the industry is currently transitioning from "thinking" to "action" in the journey towards Artificial General Intelligence (AGI) [1]. - The focus of AI competition is shifting from "model capability" to "implementation capability," with a change in value positioning from "tool efficiency" to "result closure" [2][3]. - Honor plans to address the challenges of AI implementation by connecting scenarios, performance, and trust in a closed loop, promoting collaboration among AI model companies, operators, and internet enterprises [3]. Group 2: Product Launch and Innovations - Honor announced the upcoming launch of the Magic V5 foldable flagship smartphone on July 2, which will feature multiple AI capabilities and aims to be the lightest and thinnest in its category [1][3]. - The MagicPad 3 tablet will integrate AI software and hardware, achieving seamless file transfer across iOS, HarmonyOS, and Android devices, marking a significant innovation in AI productivity [3]. - The company is committed to expanding its AI terminal ecosystem, which includes laptops, tablets, and smart wearables, as part of its Alpha strategy [3].

图片来源：企业供图 6月18日至20日，是2025年世界移动通信大会（上海）（以下简称2025MWC上海）时间，本届大会以"汇聚·连接·创造"为主题。 算力不足，内存普遍偏低，难以承载7B（模型参数量化大约为70亿）以上的模型，无法支撑复杂任务的推理需求；最后是信任，包括隐私和伦理问题。比 如说用户数据权限、数据加密等方案仍未形成统一的量化（标准）。只有打通场景、性能和信任这三大方面，才能够让AI真正走进生活，真正成为用户的 贴身伙伴。" 会上，荣耀CEO（首席执行官）李健围绕AI（人工智能）发表了主题演讲。对于"AGI（通用人工智能）到底何时到来，如何到来"的问题，他表示："目前 众说纷纭，有人说五年，也有人说需要二十年，我认同对AI五个阶段的划分：L1聊天者→L2推理者→L3行动者→L4创新者→L5组织者。全球AI行业已达成 共识，当前处于从L2到L3发展的过渡阶段，AI正在完成从'思考'到'行动'的关键跨越。" 最后，李健透露，荣耀将于7月2日发布AI折叠旗舰荣耀Magic V5，其具备全品牌互联互通的功能等。 他强调，AGI一定是一个长期演进的过程，当务之急是"让AI能解决问题"。"只有让AI落地生根 ...

Core Insights - Fourth Paradigm and Baiwang Co. have signed a strategic cooperation agreement to leverage their strengths in world models and data intelligence for innovation in transaction management, financial risk control, and decision optimization [1][3] - The collaboration aims to create integrated solutions in the financial sector, enhancing data security while providing businesses with AI capabilities for digital transformation [3][4] Group 1 - The partnership will focus on deep integration solutions combining "world models" and "data intelligence" [3] - Baiwang Co. offers comprehensive digital solutions covering procurement optimization, reconciliation, electronic invoicing, smart taxation, and financing, facilitating digital transformation in the industrial internet [3] - Fourth Paradigm has been dedicated to AI for over a decade, utilizing "AI Agent + world model" technology to empower various industries and continuously improve its capabilities [4] Group 2 - The collaboration will deepen AI cooperation and strengthen integrated solutions across the financial sector and other areas, promoting intelligent operational innovation for Chinese enterprises [6] - The goal is to support the digital transformation and high-quality development of enterprises and institutions [6]

Altman进一步表示，他认为OpenAI成功的关键在于其深植的创新文化，而Meta目前的AI发展"并未达到他们自己的期望"。他坦言，自己对Meta有很多方 面是尊重的，但"并不认为他们是一家擅长创新的公司"。他补充道："在AI领域，追赶是不够的，真正的领先来自于持续的创新。" 图片来源：Unsplash Meta首席执行官扎克伯格最近掀起了一轮激进的招人攻势，试图组建一支"超级智能（superintelligence）团队"，并将其打造成Meta进军通用人工智能 （AGI）的核心力量。据报道，这支团队由前Scale AI CEO Alexandr Wang领导，办公位置就在扎克伯格身边。为吸引顶尖AI人才，Meta向OpenAI和 Google DeepMind的研究员们开出高达1亿美元的薪酬套餐，其中包括巨额签约奖金。 对于这一传闻，OpenAI CEO Sam Altman在近日与其兄弟Jack Altman的播客节目中予以证实。不过他也毫不客气地表示，Meta的挖角行动基本以失败告 终，并借机对Meta冷嘲几句。 "Meta最近给我们团队的很多人开出了非常夸张的报价，"Altman在节目中说道，"像是 ...

不过，奥特曼指出，扎克伯格的招聘工作在很大程度上并未成功。"我很高兴，至少到目前为止，我 们最优秀的团队成员中没有人决定接受这些条件。"此前就有报道称，Meta 曾试图挖走 OpenAI 的首 席 研 究 员 诺 姆 · 布 朗 （ Noam Brown ） 以 及 谷 歌 的 AI 架 构 师 科 雷 · 卡 武 克 丘 奥 卢 （ Koray Kavukcuoglu），但均以失败告终。 "这不是建立优秀文化的方式。"奥特曼表示，Meta 将重心放在为员工提供巨额薪酬方案上，而非致 力于实现 AGI 的使命。相信员工们在比较后认为，OpenAI 在实现通用人工智能（AGI）方面更有胜 算且未来可能成为更具价值的公司。 整理 | 华卫、核子可乐 最近，Meta 首席执行官马克 · 扎克伯格（Mark Zuckerberg）掀起了一场疯狂的 AI 人才争夺战。据 外媒报道，扎克伯格正在为 Meta 新成立的超级智能团队招募来自竞争对手实验室的顶级人工智能研 究人员。为了让员工加入由前 Scale AI 首席执行官 Alexandr Wang 领导的团队，Meta 向 OpenAI 和谷歌 DeepMind 的员 ...

图片来源： Scale AI Meta 的 CEO 马克·扎克伯格近期展开大规模招聘行动，试图为 Meta 新成立的超级智能团队从竞争实 验室网罗顶尖 AI 研究人员。 在播客的后半部分，山姆 ·奥特曼描述了一个可能侵蚀 Meta 旗下应用市场的 AI 驱动社交媒体信息 流。这位 OpenAI 首席执行官表示，他对于探索一种新型社交媒体应用很感兴趣——该应用将利用 AI 根据用户需求提供定制化内容推送，而非传统社交媒体默认的算法推荐信息流。 据传， Meta 为吸引 OpenAI 和 Google DeepMind 员工加入前 Scale AI 首席执行官 Alexandr Wang 领 导的团队，并提供与扎克伯格距离相近的办公位置，开出天价薪酬方案。 OpenAI 的 CEO 萨姆·阿尔特曼在与其兄弟杰克·阿尔特曼的播客中，证实了这些报道，该播客于本周 二发布 。不过这位 OpenAI 掌门人指出，扎克伯格的挖角行动基本以失败告终，期间还不忘对 Meta 补上几句嘲讽。 （ Meta ）开始向我们团队的很多人开出天价条件， " 萨姆·阿尔特曼在播客中表示， " 比如 1 亿美元 的签约奖金，每年薪酬还不止这 ...

Core Insights - Meta is actively recruiting top AI researchers, offering salaries up to $100 million to attract talent from OpenAI and Google DeepMind, but has not successfully hired any employees from OpenAI so far [1][3] - OpenAI's CEO Sam Altman believes that OpenAI employees are more focused on the potential for achieving Artificial General Intelligence (AGI) and view OpenAI as a company with greater future value compared to Meta, which he perceives as overly focused on high salaries rather than fostering an innovative culture [3] - Meta has attempted to recruit key figures from OpenAI and Google but has faced challenges in doing so, and is planning to expand its AI team significantly in the coming year [3][4] Company Strategies - Meta is investing heavily in Scale AI and recruiting several prominent AI researchers to build a successful AI superintelligence lab [3] - The company is experimenting with AI-driven social networks through its Meta AI application, although there have been user confusion and privacy concerns associated with this initiative [4] Competitive Landscape - OpenAI is expected to release an open AI model, which may increase competitive pressure on Meta [3] - Altman mentioned that OpenAI is exploring the development of an AI-driven social media application, posing a potential threat to Meta's market position [3]