GUIRoboTron-Speech团队 投稿 量子位 | 公众号 QbitAI 只需要动动嘴就可以驱动GUI代理？ 由美团和浙江大学联合推出的 GUIRoboTron-Speech ——让用户解放双手，直接对计算机"发号施令"。 从文本到语音，智能代理的下一次进化 当前，以大型语言模型（LLMs）为核心的自主GUI智能体，已能通过文本指令自动执行跨应用、多步骤的复杂任务，极大地提升了用户的工 作效率。但这种对文本的依赖，限制了其在更广泛场景下的应用。 试想一个常见的家庭场景：在对家中的公用电脑发出指令"打开我的浏览器"时，一个仅能理解文本的智能体将不知所措——它无法分辨指令发 出者是家庭中的哪一位成员，自然不知道什么是"我的"浏览器。 然而，一个能够直接处理语音的智能体，则可以通过分析独特的声纹特征，准确识别指令发出者的身份，并打开该成员的个性化Google浏览 器界面。 这正是语音模态所蕴含的独特价值——它不仅传递了指令内容，更包含了身份、情绪等丰富的非言语线索，而这些对于实现真正个性化和智能 化的交互至关重要。 传统的解决方案，如采用"语音识别(ASR)模型转录+文本GUI代理"的级联方式，不仅会增加系 ...

演讲中，李君深入阐释了智能体作为人工智能关键演进方向的重要性。他回溯阿兰·图灵"高度智能有机体"概念，强 调智能体通过感知、规划与执行能力，实现复杂任务自主完成。"大模型是'顾问'，提供智慧大脑；智能体则成为'AI专 班'，为AI装上眼、手、脚。"他以"把大象放进冰箱"比喻，说明其拆解任务、自主执行的本质，标志AI从"说到"迈向"做 到"的关键进化。 在探讨智能体如何深刻变革传播及更广泛领域时，实验室集中探索其在传媒业的创新应用。李君指出，从AI顾问至 AI助理再到AI专班，智能体不仅赋能内容生产，还推动采编流程的全智能化，而"新闻+智能体"的创新应用，还助力主 流媒体与各行各业实现全新连接服务，逐步创造行业新生态。 李君出席代理式人工智能峰会并发表主旨演讲。人民网 严小晶摄 此外，李君重点介绍人民网"初芯"智能体平台的核心价值与实践成果。"初芯"依托人民网独有的海量优质数据资源 和各垂类AI服务能力，深度赋能人民网自身及衍生平台的内容生产、舆情研判、社会治理等全链条业务。 人民网上海6月19日电 （严小晶）2025年世界移动通信大会（上海）（简称"MWC上海"）6月18日至20日举行。人 民日报社传播内容认 ...

文｜号外工作室 资本市场即将迎来"大模型除幻第一股"！ 2025年6月17日，北京海致科技集团股份有限公司（以下简称"海致科技"）向港交所递交招股书，拟港股主板上市，此次IPO的 联席保荐人分别为招银国际、中银国际、申万宏源。 海致科技是由互联网老兵任旭阳创立的人工智能企业，他领导的这家公司用知识图谱给大模型"治病"，成为中国产业智能体赛 道隐形冠军。当资本市场迎来"大模型除幻第一股"，技术情怀与商业现实如何共舞？ 1、百度系老兵的第三次创业 海致科技成立于2013年，由百度元老、前副总裁任旭阳创立。任旭阳曾主导爱奇艺和一点资讯的创立，并担任真知创投董事 长，其丰富的互联网创业经验为海致科技注入了战略视野与资源整合能力。 2013年创立海致科技时，任旭阳将技术情怀注入公司基因："一定要做出属于中国自己的世界级软件。"如今担任董事长的他， 在招股书中被描述为"负责提供战略方向与指引"的关键角色。 海致科技的核心管理层的配置凸显技术底色：CEO杨再飞、CTO杨娟、副总万澎江组成的铁三角，平均年龄不足45岁。新加入 的CFO孙君博曾任出门问问CFO，操盘过AI公司上市全程，补齐资本运作关键拼图。 海致科技致力于通过 ...

Core Insights - The article highlights the transformative impact of AI technology on industries, particularly through the use of commercial intelligent agents like JD Cloud's JoyAgent during the 618 shopping festival [1][2]. Group 1: Commercial Intelligent Agents - The concept of "intelligent agents" is rapidly evolving, with commercial intelligent agents designed specifically for enterprise applications, distinguishing them from general-purpose agents [2]. - JoyAgent integrates deeply into the production chain, allowing employees to focus on strategic goals while the intelligent agent handles execution, creating a new productivity paradigm [2][3]. - JoyAgent's capabilities were demonstrated during the 618 event, where it analyzed vast data streams to predict inventory shortages and generated urgent replenishment plans, significantly reducing procurement time from days to minutes [2][3]. Group 2: Technological Foundation - JD's unique technology matrix, which includes a knowledge base across various sectors and a mixed-agent model, enhances JoyAgent's task handling capabilities, surpassing competitors in benchmark tests [3]. - The integration of JoyAgent into business systems allows for seamless human-machine collaboration, positioning it as a core driver of organizational evolution [3][4]. Group 3: Digital Employees - The emergence of digital employees, exemplified by JoyAgent, is reshaping organizational structures into agile networks that do not rely on traditional hierarchies [4]. - Digital employees are becoming integral to key business functions, enabling human workers to focus on higher-value tasks, thus transforming the value creation logic within organizations [4][5]. - The collaboration between humans and digital employees is likened to a continuous intelligent spiral, merging human creativity with machine execution [4][5]. Group 4: Future Competitiveness - The article posits that the future competitive advantage for companies will hinge on their ability to collaborate effectively with digital employees rather than solely on scale or manpower [5][6]. - The evolution of organizations from physical entities to entities woven from human intelligence and computational power is emphasized, with JoyAgent serving as a testament to this shift [6].

6月19日，2025上海世界移动通信大会(MWC上海)期间，荣耀CEO李健在主题演讲中宣布，将于7月2日 发布最新一代AI折叠旗舰荣耀Magic V5。据悉，全栈式个人知识库、多智能体协同带来的PC级生产 力、全品牌互联互通等都将落地到这款手机中。他表示，荣耀Magic V5将成为全球最轻薄的折叠屏手 机，也是"行业最强"AI智能体手机。(新浪财经) ...

Core Viewpoint - The article highlights the journey of Wu Yi, a prominent figure in the AI field, emphasizing his contributions to reinforcement learning and the development of open-source systems like AReaL, which aims to enhance reasoning capabilities in AI models [1][6][19]. Group 1: Wu Yi's Background and Career - Wu Yi, born in 1992, excelled in computer science competitions and was mentored by renowned professors at Tsinghua University and UC Berkeley, leading to significant internships at Microsoft and Facebook [2][4]. - After completing his PhD at UC Berkeley, Wu joined OpenAI, where he contributed to notable projects, including the "multi-agent hide-and-seek" experiment, which showcased complex behaviors emerging from simple rules [4][5]. - In 2020, Wu returned to China to teach at Tsinghua University, focusing on integrating cutting-edge technology into education and research while exploring industrial applications [5][6]. Group 2: AReaL and Reinforcement Learning - AReaL, developed in collaboration with Ant Group, is an open-source reinforcement learning framework designed to enhance reasoning models, providing efficient and reusable training solutions [6][19]. - The framework addresses the need for models to "think" before generating answers, a concept that has gained traction in recent AI developments [19][20]. - AReaL differs from traditional RLHF (Reinforcement Learning from Human Feedback) by focusing on improving the intelligence of models rather than merely making them compliant with human expectations [21][22]. Group 3: Challenges in AI Development - Wu Yi discusses the significant challenges in entrepreneurship within the AI sector, emphasizing the critical nature of timing and the risks associated with missing key opportunities [12][13]. - The evolution of model sizes presents new challenges for reinforcement learning, as modern models can have billions of parameters, necessitating adaptations in training and inference processes [23][24]. - The article also highlights the importance of data quality and system efficiency in training reinforcement learning models, asserting that these factors are more critical than algorithmic advancements [30][32]. Group 4: Future Directions in AI - Wu Yi expresses optimism about future breakthroughs in AI, particularly in areas like memory expression and personalization, which remain underexplored [40][41]. - The article suggests that while multi-agent systems are valuable, they may not be essential for all tasks, as advancements in single models could render multi-agent approaches unnecessary [42][43]. - The ongoing pursuit of scaling laws in AI development indicates that improvements in model performance will continue to be a focal point for researchers and developers [26][41].

Group 1: Federal Reserve and Economic Outlook - The Federal Reserve maintains the federal funds rate target range at 4.25%-4.5% and anticipates two rate cuts by the end of the year [2] - Economic growth forecast for this year has been downgraded to 1.4%, while inflation expectations have been raised to 3% [2] - The labor market remains strong, with no signs of economic weakness, but uncertainties regarding trade and fiscal policies persist [2][4] Group 2: AI and Internet Sector Insights - UBS reports that the KWEB China Internet ETF has risen 18% year-to-date, driven by valuation, particularly in AI stocks [5] - Key focus areas for the second half of the year include AI monetization, overseas expansion, and profit margin restructuring [5] - The transition from commission to advertising revenue is expected to enhance profit margins for e-commerce platforms [5] Group 3: Global Market Sentiment - A Bank of America survey indicates that 54% of fund managers favor international stocks over U.S. stocks for the next five years [8] - Concerns about trade wars and potential global recession are highlighted as significant tail risks [8] - Investor sentiment has improved, with 66% believing in a soft landing for the global economy in the next 12 months [8] Group 4: AI Transformation in Software Industry - Goldman Sachs predicts that "intelligent AI" will transform the enterprise software ecosystem, with a market size expected to grow by at least 20% by 2030 [10] - The customer service software market is projected to grow at a rate of 45%, with intelligent AI expected to capture over 60% of the software industry [10] - Companies like Microsoft, Google, and Adobe are recommended for investment due to their potential in the new AI ecosystem [10] Group 5: Gene Editing Sector Developments - Eli Lilly's acquisition of Verve Therapeutics for up to $1.3 billion signals a positive outlook for the gene editing industry [11] - Verve's stock surged by 81.5% following the acquisition announcement, indicating strong market interest in gene therapy [11] - The investment logic in gene editing is shifting towards specific targets and clear payment models, moving beyond platform potential [12]

Group 1 - Goldman Sachs reports that the next phase of generative AI, termed "Agentic AI," will significantly transform the enterprise software ecosystem [1][2] - Over the next three years, Agentic AI is expected to unlock productivity gains at the application layer, with the global software market projected to expand by at least 20% by 2030 [2][3] - The customer service software market could see growth rates between 20% to 45%, driven by the integration of traditional SaaS and AI agents [2][3] Group 2 - SaaS companies are anticipated to capture a substantial share of the new Agentic AI market, but their innovation pace is critical, and the transition may not be linear [3][4] - By 2030, Agentic AI is expected to account for over 60% of the total software market, potentially becoming the new user interface for knowledge workers [3][4] - Existing SaaS leaders are showing signs of enhancing execution capabilities, indicating a clear strategic market awareness [3][4] Group 3 - The technological architecture for generative AI applications will require a new tech stack, leading to significant changes in existing architectures [4] - The rise of AI platform layers and the improvement of key middleware will be crucial for the development of AI-native applications [4] - SaaS companies must adapt to emerging AI standards and adjust their architectures to successfully integrate into the generative AI enterprise application ecosystem [4][5] Group 4 - Despite current limitations in SaaS giants' transitions due to generative AI technology maturity, these factors are expected to translate into sustained growth momentum after 2027 [5] - Investors are advised to focus on companies such as Microsoft, Google, Salesforce, ServiceNow, HubSpot, Adobe, and several private firms as potential investment opportunities [5]

Core Insights - The rise of intelligent agents in the AI sector is being driven by significant investments from various Chinese tech companies, with predictions that 2025 may mark a breakthrough year for this technology [1][2] - Intelligent agents are defined as interactive systems that utilize AI to understand external stimuli and generate meaningful actions, encompassing key technologies such as environmental perception, decision planning, autonomous learning, multimodal interaction, and task execution [2] Company Developments - Luckin Coffee has implemented an intelligent agent in its app, allowing users to order coffee through voice commands, significantly enhancing order efficiency [1] - Lenovo has launched the Tianxi personal super intelligent agent, which integrates into personal computers, smartphones, and tablets, focusing on multimodal interaction while ensuring data privacy and security [1] - JD.com operates over 14,000 intelligent agents that handle more than 18% of work tasks, particularly in areas like food delivery recruitment and financial management [1] Industry Trends - The development of intelligent agents is seen as a catalyst for industrial upgrades, creating new business models and economic growth points [2] - The technology is expected to evolve from passive tools to proactive executors, with the potential for widespread application in daily life and work environments [3] - Experts suggest that specialized intelligent agents may have a greater chance of successful implementation compared to general-purpose agents, with the potential for significant growth in the SaaS market [3]

辅导老师介绍 老师均在CVPR、ICCV、ECCV、ICLR、RSS、ICML、ICRA等顶级会议上发表论文，有较丰富的 指导经验。 学员要求 自带一份简历，学校背景：国内TOP100高校，国外QS200以内； 具身智能之心论文辅导正式推出啦！去年的成果还算不错，几个同学中了CVPR和ICRA等会议， 今年和老师们沟通过后，准备继续辅导几名同学冲下顶会，感兴趣的同学可以咨询，辅导方向如 下。 主要方向 更多咨询 多模态大模型，VLA、机器人导航、机器人抓取、具身泛化、具身合成数据、端到端具身智能 体、3DGS等方向； 详细内容欢迎添加微信：oooops-life，做进一步了解。 ...